qemu/hw/usb-uhci.c

1215 lines
31 KiB
C
Raw Normal View History

/*
* USB UHCI controller emulation
*
* Copyright (c) 2005 Fabrice Bellard
*
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
* Copyright (c) 2008 Max Krasnyansky
* Magor rewrite of the UHCI data structures parser and frame processor
* Support for fully async operation and multiple outstanding transactions
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include "hw.h"
#include "usb.h"
#include "pci.h"
#include "qemu-timer.h"
#include "usb-uhci.h"
//#define DEBUG
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
//#define DEBUG_DUMP_DATA
#define UHCI_CMD_FGR (1 << 4)
#define UHCI_CMD_EGSM (1 << 3)
#define UHCI_CMD_GRESET (1 << 2)
#define UHCI_CMD_HCRESET (1 << 1)
#define UHCI_CMD_RS (1 << 0)
#define UHCI_STS_HCHALTED (1 << 5)
#define UHCI_STS_HCPERR (1 << 4)
#define UHCI_STS_HSERR (1 << 3)
#define UHCI_STS_RD (1 << 2)
#define UHCI_STS_USBERR (1 << 1)
#define UHCI_STS_USBINT (1 << 0)
#define TD_CTRL_SPD (1 << 29)
#define TD_CTRL_ERROR_SHIFT 27
#define TD_CTRL_IOS (1 << 25)
#define TD_CTRL_IOC (1 << 24)
#define TD_CTRL_ACTIVE (1 << 23)
#define TD_CTRL_STALL (1 << 22)
#define TD_CTRL_BABBLE (1 << 20)
#define TD_CTRL_NAK (1 << 19)
#define TD_CTRL_TIMEOUT (1 << 18)
#define UHCI_PORT_RESET (1 << 9)
#define UHCI_PORT_LSDA (1 << 8)
#define UHCI_PORT_ENC (1 << 3)
#define UHCI_PORT_EN (1 << 2)
#define UHCI_PORT_CSC (1 << 1)
#define UHCI_PORT_CCS (1 << 0)
#define FRAME_TIMER_FREQ 1000
#define FRAME_MAX_LOOPS 100
#define NB_PORTS 2
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
#ifdef DEBUG
#define DPRINTF printf
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
static const char *pid2str(int pid)
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
{
switch (pid) {
case USB_TOKEN_SETUP: return "SETUP";
case USB_TOKEN_IN: return "IN";
case USB_TOKEN_OUT: return "OUT";
}
return "?";
}
#else
#define DPRINTF(...)
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
#endif
#ifdef DEBUG_DUMP_DATA
static void dump_data(const uint8_t *data, int len)
{
int i;
printf("uhci: data: ");
for(i = 0; i < len; i++)
printf(" %02x", data[i]);
printf("\n");
}
#else
static void dump_data(const uint8_t *data, int len) {}
#endif
/*
* Pending async transaction.
* 'packet' must be the first field because completion
* handler does "(UHCIAsync *) pkt" cast.
*/
typedef struct UHCIAsync {
USBPacket packet;
struct UHCIAsync *next;
uint32_t td;
uint32_t token;
int8_t valid;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
uint8_t isoc;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
uint8_t done;
uint8_t buffer[2048];
} UHCIAsync;
typedef struct UHCIPort {
USBPort port;
uint16_t ctrl;
} UHCIPort;
typedef struct UHCIState {
PCIDevice dev;
USBBus bus;
uint16_t cmd; /* cmd register */
uint16_t status;
uint16_t intr; /* interrupt enable register */
uint16_t frnum; /* frame number */
uint32_t fl_base_addr; /* frame list base address */
uint8_t sof_timing;
uint8_t status2; /* bit 0 and 1 are used to generate UHCI_STS_USBINT */
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
int64_t expire_time;
QEMUTimer *frame_timer;
UHCIPort ports[NB_PORTS];
/* Interrupts that should be raised at the end of the current frame. */
uint32_t pending_int_mask;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* Active packets */
UHCIAsync *async_pending;
UHCIAsync *async_pool;
uint8_t num_ports_vmstate;
} UHCIState;
typedef struct UHCI_TD {
uint32_t link;
uint32_t ctrl; /* see TD_CTRL_xxx */
uint32_t token;
uint32_t buffer;
} UHCI_TD;
typedef struct UHCI_QH {
uint32_t link;
uint32_t el_link;
} UHCI_QH;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
static UHCIAsync *uhci_async_alloc(UHCIState *s)
{
UHCIAsync *async = qemu_malloc(sizeof(UHCIAsync));
memset(&async->packet, 0, sizeof(async->packet));
async->valid = 0;
async->td = 0;
async->token = 0;
async->done = 0;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
async->isoc = 0;
async->next = NULL;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
return async;
}
static void uhci_async_free(UHCIState *s, UHCIAsync *async)
{
qemu_free(async);
}
static void uhci_async_link(UHCIState *s, UHCIAsync *async)
{
async->next = s->async_pending;
s->async_pending = async;
}
static void uhci_async_unlink(UHCIState *s, UHCIAsync *async)
{
UHCIAsync *curr = s->async_pending;
UHCIAsync **prev = &s->async_pending;
while (curr) {
if (curr == async) {
*prev = curr->next;
return;
}
prev = &curr->next;
curr = curr->next;
}
}
static void uhci_async_cancel(UHCIState *s, UHCIAsync *async)
{
DPRINTF("uhci: cancel td 0x%x token 0x%x done %u\n",
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
async->td, async->token, async->done);
if (!async->done)
usb_cancel_packet(&async->packet);
uhci_async_free(s, async);
}
/*
* Mark all outstanding async packets as invalid.
* This is used for canceling them when TDs are removed by the HCD.
*/
static UHCIAsync *uhci_async_validate_begin(UHCIState *s)
{
UHCIAsync *async = s->async_pending;
while (async) {
async->valid--;
async = async->next;
}
return NULL;
}
/*
* Cancel async packets that are no longer valid
*/
static void uhci_async_validate_end(UHCIState *s)
{
UHCIAsync *curr = s->async_pending;
UHCIAsync **prev = &s->async_pending;
UHCIAsync *next;
while (curr) {
if (curr->valid > 0) {
prev = &curr->next;
curr = curr->next;
continue;
}
next = curr->next;
/* Unlink */
*prev = next;
uhci_async_cancel(s, curr);
curr = next;
}
}
static void uhci_async_cancel_all(UHCIState *s)
{
UHCIAsync *curr = s->async_pending;
UHCIAsync *next;
while (curr) {
next = curr->next;
uhci_async_cancel(s, curr);
curr = next;
}
s->async_pending = NULL;
}
static UHCIAsync *uhci_async_find_td(UHCIState *s, uint32_t addr, uint32_t token)
{
UHCIAsync *async = s->async_pending;
UHCIAsync *match = NULL;
int count = 0;
/*
* We're looking for the best match here. ie both td addr and token.
* Otherwise we return last good match. ie just token.
* It's ok to match just token because it identifies the transaction
* rather well, token includes: device addr, endpoint, size, etc.
*
* Also since we queue async transactions in reverse order by returning
* last good match we restores the order.
*
* It's expected that we wont have a ton of outstanding transactions.
* If we ever do we'd want to optimize this algorithm.
*/
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
while (async) {
if (async->token == token) {
/* Good match */
match = async;
if (async->td == addr) {
/* Best match */
break;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
}
}
async = async->next;
count++;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
}
if (count > 64)
fprintf(stderr, "uhci: warning lots of async transactions\n");
return match;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
}
static void uhci_attach(USBPort *port1, USBDevice *dev);
static void uhci_update_irq(UHCIState *s)
{
int level;
if (((s->status2 & 1) && (s->intr & (1 << 2))) ||
((s->status2 & 2) && (s->intr & (1 << 3))) ||
((s->status & UHCI_STS_USBERR) && (s->intr & (1 << 0))) ||
((s->status & UHCI_STS_RD) && (s->intr & (1 << 1))) ||
(s->status & UHCI_STS_HSERR) ||
(s->status & UHCI_STS_HCPERR)) {
level = 1;
} else {
level = 0;
}
qemu_set_irq(s->dev.irq[3], level);
}
static void uhci_reset(void *opaque)
{
UHCIState *s = opaque;
uint8_t *pci_conf;
int i;
UHCIPort *port;
DPRINTF("uhci: full reset\n");
pci_conf = s->dev.config;
pci_conf[0x6a] = 0x01; /* usb clock */
pci_conf[0x6b] = 0x00;
s->cmd = 0;
s->status = 0;
s->status2 = 0;
s->intr = 0;
s->fl_base_addr = 0;
s->sof_timing = 64;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
for(i = 0; i < NB_PORTS; i++) {
port = &s->ports[i];
port->ctrl = 0x0080;
if (port->port.dev)
uhci_attach(&port->port, port->port.dev);
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
uhci_async_cancel_all(s);
}
static void uhci_pre_save(void *opaque)
{
UHCIState *s = opaque;
uhci_async_cancel_all(s);
}
static const VMStateDescription vmstate_uhci_port = {
.name = "uhci port",
.version_id = 1,
.minimum_version_id = 1,
.minimum_version_id_old = 1,
.fields = (VMStateField []) {
VMSTATE_UINT16(ctrl, UHCIPort),
VMSTATE_END_OF_LIST()
}
};
static const VMStateDescription vmstate_uhci = {
.name = "uhci",
.version_id = 2,
.minimum_version_id = 1,
.minimum_version_id_old = 1,
.pre_save = uhci_pre_save,
.fields = (VMStateField []) {
VMSTATE_PCI_DEVICE(dev, UHCIState),
VMSTATE_UINT8_EQUAL(num_ports_vmstate, UHCIState),
VMSTATE_STRUCT_ARRAY(ports, UHCIState, NB_PORTS, 1,
vmstate_uhci_port, UHCIPort),
VMSTATE_UINT16(cmd, UHCIState),
VMSTATE_UINT16(status, UHCIState),
VMSTATE_UINT16(intr, UHCIState),
VMSTATE_UINT16(frnum, UHCIState),
VMSTATE_UINT32(fl_base_addr, UHCIState),
VMSTATE_UINT8(sof_timing, UHCIState),
VMSTATE_UINT8(status2, UHCIState),
VMSTATE_TIMER(frame_timer, UHCIState),
VMSTATE_INT64_V(expire_time, UHCIState, 2),
VMSTATE_END_OF_LIST()
}
};
static void uhci_ioport_writeb(void *opaque, uint32_t addr, uint32_t val)
{
UHCIState *s = opaque;
addr &= 0x1f;
switch(addr) {
case 0x0c:
s->sof_timing = val;
break;
}
}
static uint32_t uhci_ioport_readb(void *opaque, uint32_t addr)
{
UHCIState *s = opaque;
uint32_t val;
addr &= 0x1f;
switch(addr) {
case 0x0c:
val = s->sof_timing;
break;
default:
val = 0xff;
break;
}
return val;
}
static void uhci_ioport_writew(void *opaque, uint32_t addr, uint32_t val)
{
UHCIState *s = opaque;
addr &= 0x1f;
DPRINTF("uhci: writew port=0x%04x val=0x%04x\n", addr, val);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
switch(addr) {
case 0x00:
if ((val & UHCI_CMD_RS) && !(s->cmd & UHCI_CMD_RS)) {
/* start frame processing */
qemu_mod_timer(s->frame_timer, qemu_get_clock(vm_clock));
s->status &= ~UHCI_STS_HCHALTED;
} else if (!(val & UHCI_CMD_RS)) {
s->status |= UHCI_STS_HCHALTED;
}
if (val & UHCI_CMD_GRESET) {
UHCIPort *port;
USBDevice *dev;
int i;
/* send reset on the USB bus */
for(i = 0; i < NB_PORTS; i++) {
port = &s->ports[i];
dev = port->port.dev;
if (dev) {
usb_send_msg(dev, USB_MSG_RESET);
}
}
uhci_reset(s);
return;
}
if (val & UHCI_CMD_HCRESET) {
uhci_reset(s);
return;
}
s->cmd = val;
break;
case 0x02:
s->status &= ~val;
/* XXX: the chip spec is not coherent, so we add a hidden
register to distinguish between IOC and SPD */
if (val & UHCI_STS_USBINT)
s->status2 = 0;
uhci_update_irq(s);
break;
case 0x04:
s->intr = val;
uhci_update_irq(s);
break;
case 0x06:
if (s->status & UHCI_STS_HCHALTED)
s->frnum = val & 0x7ff;
break;
case 0x10 ... 0x1f:
{
UHCIPort *port;
USBDevice *dev;
int n;
n = (addr >> 1) & 7;
if (n >= NB_PORTS)
return;
port = &s->ports[n];
dev = port->port.dev;
if (dev) {
/* port reset */
if ( (val & UHCI_PORT_RESET) &&
!(port->ctrl & UHCI_PORT_RESET) ) {
usb_send_msg(dev, USB_MSG_RESET);
}
}
port->ctrl = (port->ctrl & 0x01fb) | (val & ~0x01fb);
/* some bits are reset when a '1' is written to them */
port->ctrl &= ~(val & 0x000a);
}
break;
}
}
static uint32_t uhci_ioport_readw(void *opaque, uint32_t addr)
{
UHCIState *s = opaque;
uint32_t val;
addr &= 0x1f;
switch(addr) {
case 0x00:
val = s->cmd;
break;
case 0x02:
val = s->status;
break;
case 0x04:
val = s->intr;
break;
case 0x06:
val = s->frnum;
break;
case 0x10 ... 0x1f:
{
UHCIPort *port;
int n;
n = (addr >> 1) & 7;
if (n >= NB_PORTS)
goto read_default;
port = &s->ports[n];
val = port->ctrl;
}
break;
default:
read_default:
val = 0xff7f; /* disabled port */
break;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
DPRINTF("uhci: readw port=0x%04x val=0x%04x\n", addr, val);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
return val;
}
static void uhci_ioport_writel(void *opaque, uint32_t addr, uint32_t val)
{
UHCIState *s = opaque;
addr &= 0x1f;
DPRINTF("uhci: writel port=0x%04x val=0x%08x\n", addr, val);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
switch(addr) {
case 0x08:
s->fl_base_addr = val & ~0xfff;
break;
}
}
static uint32_t uhci_ioport_readl(void *opaque, uint32_t addr)
{
UHCIState *s = opaque;
uint32_t val;
addr &= 0x1f;
switch(addr) {
case 0x08:
val = s->fl_base_addr;
break;
default:
val = 0xffffffff;
break;
}
return val;
}
/* signal resume if controller suspended */
static void uhci_resume (void *opaque)
{
UHCIState *s = (UHCIState *)opaque;
if (!s)
return;
if (s->cmd & UHCI_CMD_EGSM) {
s->cmd |= UHCI_CMD_FGR;
s->status |= UHCI_STS_RD;
uhci_update_irq(s);
}
}
static void uhci_attach(USBPort *port1, USBDevice *dev)
{
UHCIState *s = port1->opaque;
UHCIPort *port = &s->ports[port1->index];
if (dev) {
if (port->port.dev) {
usb_attach(port1, NULL);
}
/* set connect status */
port->ctrl |= UHCI_PORT_CCS | UHCI_PORT_CSC;
/* update speed */
if (dev->speed == USB_SPEED_LOW)
port->ctrl |= UHCI_PORT_LSDA;
else
port->ctrl &= ~UHCI_PORT_LSDA;
uhci_resume(s);
port->port.dev = dev;
/* send the attach message */
usb_send_msg(dev, USB_MSG_ATTACH);
} else {
/* set connect status */
if (port->ctrl & UHCI_PORT_CCS) {
port->ctrl &= ~UHCI_PORT_CCS;
port->ctrl |= UHCI_PORT_CSC;
}
/* disable port */
if (port->ctrl & UHCI_PORT_EN) {
port->ctrl &= ~UHCI_PORT_EN;
port->ctrl |= UHCI_PORT_ENC;
}
uhci_resume(s);
dev = port->port.dev;
if (dev) {
/* send the detach message */
usb_send_msg(dev, USB_MSG_DETACH);
}
port->port.dev = NULL;
}
}
static int uhci_broadcast_packet(UHCIState *s, USBPacket *p)
{
int i, ret;
DPRINTF("uhci: packet enter. pid %s addr 0x%02x ep %d len %d\n",
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
pid2str(p->pid), p->devaddr, p->devep, p->len);
if (p->pid == USB_TOKEN_OUT || p->pid == USB_TOKEN_SETUP)
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
dump_data(p->data, p->len);
ret = USB_RET_NODEV;
for (i = 0; i < NB_PORTS && ret == USB_RET_NODEV; i++) {
UHCIPort *port = &s->ports[i];
USBDevice *dev = port->port.dev;
if (dev && (port->ctrl & UHCI_PORT_EN))
ret = dev->info->handle_packet(dev, p);
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
DPRINTF("uhci: packet exit. ret %d len %d\n", ret, p->len);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (p->pid == USB_TOKEN_IN && ret > 0)
dump_data(p->data, ret);
return ret;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
static void uhci_async_complete(USBPacket * packet, void *opaque);
static void uhci_process_frame(UHCIState *s);
/* return -1 if fatal error (frame must be stopped)
0 if TD successful
1 if TD unsuccessful or inactive
*/
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
static int uhci_complete_td(UHCIState *s, UHCI_TD *td, UHCIAsync *async, uint32_t *int_mask)
{
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
int len = 0, max_len, err, ret;
uint8_t pid;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
max_len = ((td->token >> 21) + 1) & 0x7ff;
pid = td->token & 0xff;
ret = async->packet.len;
if (td->ctrl & TD_CTRL_IOS)
td->ctrl &= ~TD_CTRL_ACTIVE;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (ret < 0)
goto out;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
len = async->packet.len;
td->ctrl = (td->ctrl & ~0x7ff) | ((len - 1) & 0x7ff);
/* The NAK bit may have been set by a previous frame, so clear it
here. The docs are somewhat unclear, but win2k relies on this
behavior. */
td->ctrl &= ~(TD_CTRL_ACTIVE | TD_CTRL_NAK);
if (td->ctrl & TD_CTRL_IOC)
*int_mask |= 0x01;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (pid == USB_TOKEN_IN) {
if (len > max_len) {
ret = USB_RET_BABBLE;
goto out;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (len > 0) {
/* write the data back */
cpu_physical_memory_write(td->buffer, async->buffer, len);
}
if ((td->ctrl & TD_CTRL_SPD) && len < max_len) {
*int_mask |= 0x02;
/* short packet: do not update QH */
DPRINTF("uhci: short packet. td 0x%x token 0x%x\n", async->td, async->token);
return 1;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
}
/* success */
return 0;
out:
switch(ret) {
case USB_RET_STALL:
td->ctrl |= TD_CTRL_STALL;
td->ctrl &= ~TD_CTRL_ACTIVE;
return 1;
case USB_RET_BABBLE:
td->ctrl |= TD_CTRL_BABBLE | TD_CTRL_STALL;
td->ctrl &= ~TD_CTRL_ACTIVE;
/* frame interrupted */
return -1;
case USB_RET_NAK:
td->ctrl |= TD_CTRL_NAK;
if (pid == USB_TOKEN_SETUP)
break;
return 1;
case USB_RET_NODEV:
default:
break;
}
/* Retry the TD if error count is not zero */
td->ctrl |= TD_CTRL_TIMEOUT;
err = (td->ctrl >> TD_CTRL_ERROR_SHIFT) & 3;
if (err != 0) {
err--;
if (err == 0) {
td->ctrl &= ~TD_CTRL_ACTIVE;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
s->status |= UHCI_STS_USBERR;
if (td->ctrl & TD_CTRL_IOC)
*int_mask |= 0x01;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
uhci_update_irq(s);
}
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
td->ctrl = (td->ctrl & ~(3 << TD_CTRL_ERROR_SHIFT)) |
(err << TD_CTRL_ERROR_SHIFT);
return 1;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
static int uhci_handle_td(UHCIState *s, uint32_t addr, UHCI_TD *td, uint32_t *int_mask)
{
UHCIAsync *async;
int len = 0, max_len;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
uint8_t pid, isoc;
uint32_t token;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* Is active ? */
if (!(td->ctrl & TD_CTRL_ACTIVE))
return 1;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
/* token field is not unique for isochronous requests,
* so use the destination buffer
*/
if (td->ctrl & TD_CTRL_IOS) {
token = td->buffer;
isoc = 1;
} else {
token = td->token;
isoc = 0;
}
async = uhci_async_find_td(s, addr, token);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (async) {
/* Already submitted */
async->valid = 32;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (!async->done)
return 1;
uhci_async_unlink(s, async);
goto done;
}
/* Allocate new packet */
async = uhci_async_alloc(s);
if (!async)
return 1;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
/* valid needs to be large enough to handle 10 frame delay
* for initial isochronous requests
*/
async->valid = 32;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
async->td = addr;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
async->token = token;
async->isoc = isoc;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
max_len = ((td->token >> 21) + 1) & 0x7ff;
pid = td->token & 0xff;
async->packet.pid = pid;
async->packet.devaddr = (td->token >> 8) & 0x7f;
async->packet.devep = (td->token >> 15) & 0xf;
async->packet.data = async->buffer;
async->packet.len = max_len;
async->packet.complete_cb = uhci_async_complete;
async->packet.complete_opaque = s;
switch(pid) {
case USB_TOKEN_OUT:
case USB_TOKEN_SETUP:
cpu_physical_memory_read(td->buffer, async->buffer, max_len);
len = uhci_broadcast_packet(s, &async->packet);
if (len >= 0)
len = max_len;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
break;
case USB_TOKEN_IN:
len = uhci_broadcast_packet(s, &async->packet);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
break;
default:
/* invalid pid : frame interrupted */
uhci_async_free(s, async);
s->status |= UHCI_STS_HCPERR;
uhci_update_irq(s);
return -1;
}
if (len == USB_RET_ASYNC) {
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
uhci_async_link(s, async);
return 2;
}
async->packet.len = len;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
done:
len = uhci_complete_td(s, td, async, int_mask);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
uhci_async_free(s, async);
return len;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
}
static void uhci_async_complete(USBPacket *packet, void *opaque)
{
UHCIState *s = opaque;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
UHCIAsync *async = (UHCIAsync *) packet;
DPRINTF("uhci: async complete. td 0x%x token 0x%x\n", async->td, async->token);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
if (async->isoc) {
UHCI_TD td;
uint32_t link = async->td;
uint32_t int_mask = 0, val;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
cpu_physical_memory_read(link & ~0xf, (uint8_t *) &td, sizeof(td));
le32_to_cpus(&td.link);
le32_to_cpus(&td.ctrl);
le32_to_cpus(&td.token);
le32_to_cpus(&td.buffer);
uhci_async_unlink(s, async);
uhci_complete_td(s, &td, async, &int_mask);
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
s->pending_int_mask |= int_mask;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
/* update the status bits of the TD */
val = cpu_to_le32(td.ctrl);
cpu_physical_memory_write((link & ~0xf) + 4,
(const uint8_t *)&val, sizeof(val));
uhci_async_free(s, async);
} else {
async->done = 1;
uhci_process_frame(s);
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
}
static int is_valid(uint32_t link)
{
return (link & 1) == 0;
}
static int is_qh(uint32_t link)
{
return (link & 2) != 0;
}
static int depth_first(uint32_t link)
{
return (link & 4) != 0;
}
/* QH DB used for detecting QH loops */
#define UHCI_MAX_QUEUES 128
typedef struct {
uint32_t addr[UHCI_MAX_QUEUES];
int count;
} QhDb;
static void qhdb_reset(QhDb *db)
{
db->count = 0;
}
/* Add QH to DB. Returns 1 if already present or DB is full. */
static int qhdb_insert(QhDb *db, uint32_t addr)
{
int i;
for (i = 0; i < db->count; i++)
if (db->addr[i] == addr)
return 1;
if (db->count >= UHCI_MAX_QUEUES)
return 1;
db->addr[db->count++] = addr;
return 0;
}
static void uhci_process_frame(UHCIState *s)
{
uint32_t frame_addr, link, old_td_ctrl, val, int_mask;
uint32_t curr_qh;
int cnt, ret;
UHCI_TD td;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
UHCI_QH qh;
QhDb qhdb;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
frame_addr = s->fl_base_addr + ((s->frnum & 0x3ff) << 2);
DPRINTF("uhci: processing frame %d addr 0x%x\n" , s->frnum, frame_addr);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
cpu_physical_memory_read(frame_addr, (uint8_t *)&link, 4);
le32_to_cpus(&link);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
int_mask = 0;
curr_qh = 0;
qhdb_reset(&qhdb);
for (cnt = FRAME_MAX_LOOPS; is_valid(link) && cnt; cnt--) {
if (is_qh(link)) {
/* QH */
if (qhdb_insert(&qhdb, link)) {
/*
* We're going in circles. Which is not a bug because
* HCD is allowed to do that as part of the BW management.
* In our case though it makes no sense to spin here. Sync transations
* are already done, and async completion handler will re-process
* the frame when something is ready.
*/
DPRINTF("uhci: detected loop. qh 0x%x\n", link);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
break;
}
cpu_physical_memory_read(link & ~0xf, (uint8_t *) &qh, sizeof(qh));
le32_to_cpus(&qh.link);
le32_to_cpus(&qh.el_link);
DPRINTF("uhci: QH 0x%x load. link 0x%x elink 0x%x\n",
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
link, qh.link, qh.el_link);
if (!is_valid(qh.el_link)) {
/* QH w/o elements */
curr_qh = 0;
link = qh.link;
} else {
/* QH with elements */
curr_qh = link;
link = qh.el_link;
}
continue;
}
/* TD */
cpu_physical_memory_read(link & ~0xf, (uint8_t *) &td, sizeof(td));
le32_to_cpus(&td.link);
le32_to_cpus(&td.ctrl);
le32_to_cpus(&td.token);
le32_to_cpus(&td.buffer);
DPRINTF("uhci: TD 0x%x load. link 0x%x ctrl 0x%x token 0x%x qh 0x%x\n",
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
link, td.link, td.ctrl, td.token, curr_qh);
old_td_ctrl = td.ctrl;
ret = uhci_handle_td(s, link, &td, &int_mask);
if (old_td_ctrl != td.ctrl) {
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* update the status bits of the TD */
val = cpu_to_le32(td.ctrl);
cpu_physical_memory_write((link & ~0xf) + 4,
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
(const uint8_t *)&val, sizeof(val));
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (ret < 0) {
/* interrupted frame */
break;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
if (ret == 2 || ret == 1) {
DPRINTF("uhci: TD 0x%x %s. link 0x%x ctrl 0x%x token 0x%x qh 0x%x\n",
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
link, ret == 2 ? "pend" : "skip",
td.link, td.ctrl, td.token, curr_qh);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
link = curr_qh ? qh.link : td.link;
continue;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* completed TD */
DPRINTF("uhci: TD 0x%x done. link 0x%x ctrl 0x%x token 0x%x qh 0x%x\n",
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
link, td.link, td.ctrl, td.token, curr_qh);
link = td.link;
if (curr_qh) {
/* update QH element link */
qh.el_link = link;
val = cpu_to_le32(qh.el_link);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
cpu_physical_memory_write((curr_qh & ~0xf) + 4,
(const uint8_t *)&val, sizeof(val));
if (!depth_first(link)) {
/* done with this QH */
DPRINTF("uhci: QH 0x%x done. link 0x%x elink 0x%x\n",
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
curr_qh, qh.link, qh.el_link);
curr_qh = 0;
link = qh.link;
}
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* go to the next entry */
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
s->pending_int_mask |= int_mask;
}
static void uhci_frame_timer(void *opaque)
{
UHCIState *s = opaque;
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
/* prepare the timer for the next frame */
s->expire_time += (get_ticks_per_sec() / FRAME_TIMER_FREQ);
if (!(s->cmd & UHCI_CMD_RS)) {
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* Full stop */
qemu_del_timer(s->frame_timer);
/* set hchalted bit in status - UHCI11D 2.1.2 */
s->status |= UHCI_STS_HCHALTED;
DPRINTF("uhci: halted\n");
return;
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* Complete the previous frame */
if (s->pending_int_mask) {
s->status2 |= s->pending_int_mask;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
s->status |= UHCI_STS_USBINT;
uhci_update_irq(s);
}
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
s->pending_int_mask = 0;
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
/* Start new frame */
s->frnum = (s->frnum + 1) & 0x7ff;
DPRINTF("uhci: new frame #%u\n" , s->frnum);
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
uhci_async_validate_begin(s);
uhci_process_frame(s);
uhci_async_validate_end(s);
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
qemu_mod_timer(s->frame_timer, s->expire_time);
}
static void uhci_map(PCIDevice *pci_dev, int region_num,
pcibus_t addr, pcibus_t size, int type)
{
UHCIState *s = (UHCIState *)pci_dev;
register_ioport_write(addr, 32, 2, uhci_ioport_writew, s);
register_ioport_read(addr, 32, 2, uhci_ioport_readw, s);
register_ioport_write(addr, 32, 4, uhci_ioport_writel, s);
register_ioport_read(addr, 32, 4, uhci_ioport_readl, s);
register_ioport_write(addr, 32, 1, uhci_ioport_writeb, s);
register_ioport_read(addr, 32, 1, uhci_ioport_readb, s);
}
static int usb_uhci_common_initfn(UHCIState *s)
{
uint8_t *pci_conf = s->dev.config;
int i;
pci_conf[PCI_REVISION_ID] = 0x01; // revision number
pci_conf[PCI_CLASS_PROG] = 0x00;
pci_config_set_class(pci_conf, PCI_CLASS_SERIAL_USB);
pci_conf[PCI_HEADER_TYPE] = PCI_HEADER_TYPE_NORMAL; // header_type
/* TODO: reset value should be 0. */
pci_conf[PCI_INTERRUPT_PIN] = 4; // interrupt pin 3
pci_conf[0x60] = 0x10; // release number
usb_bus_new(&s->bus, &s->dev.qdev);
for(i = 0; i < NB_PORTS; i++) {
usb_register_port(&s->bus, &s->ports[i].port, s, i, uhci_attach);
}
s->frame_timer = qemu_new_timer(vm_clock, uhci_frame_timer, s);
audio streaming from usb devices I have streaming audio devices working within qemu-kvm. This is a port of the changes to qemu. Streaming audio generates a series of isochronous requests that are repetitive and time sensitive. The URBs need to be submitted in consecutive USB frames and responses need to be handled in a timely manner. Summary of the changes for isochronous requests: 1. The initial 'valid' value is increased to 32. It needs to be higher than its current value of 10 since the host adds a 10 frame delay to the scheduling of the first request; if valid is set to 10 the first isochronous request times out and qemu cancels it. 32 was chosen as a nice round number, and it is used in the path where a TD-async pairing already exists. 2. The token field in the TD is *not* unique for isochronous requests, so it is not a good choice for finding a matching async request. The buffer (where to write the guest data) is unique, so use that value instead. 3. TD's for isochronous request need to be completed in the async completion handler so that data is pushed to the guest as soon as it is available. The uhci code currently attempts to process complete isochronous TDs the next time the UHCI frame with the request is processed. The results in lost data since the async requests will have long since timed out based on the valid parameter. Increasing the valid value is not acceptable as it introduces a 1+ second delay in the data getting pushed to the guest. 4. The frame timer needs to be run on 1 msec intervals. Currently, the expire time for the processing the next frame is computed after the processing of each frame. This regularly causes the scheduling of frames to shift in time. When this happens the periodic scheduling of the requests is broken and the subsequent request is seen as a new request by the host resulting in a 10 msec delay (first isochronous request is scheduled for 10 frames from when the URB is submitted). [ For what's worth a small change is needed to the guest driver to have more outstanding URBs (at least 4 URBs with 5 packets per URB).] Signed-off-by: David Ahern <daahern@cisco.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2010-02-03 18:49:39 +03:00
s->expire_time = qemu_get_clock(vm_clock) +
(get_ticks_per_sec() / FRAME_TIMER_FREQ);
s->num_ports_vmstate = NB_PORTS;
qemu_register_reset(uhci_reset, s);
/* Use region 4 for consistency with real hardware. BSD guests seem
to rely on this. */
pci_register_bar(&s->dev, 4, 0x20,
PCI_BASE_ADDRESS_SPACE_IO, uhci_map);
return 0;
}
static int usb_uhci_piix3_initfn(PCIDevice *dev)
{
UHCIState *s = DO_UPCAST(UHCIState, dev, dev);
uint8_t *pci_conf = s->dev.config;
pci_config_set_vendor_id(pci_conf, PCI_VENDOR_ID_INTEL);
pci_config_set_device_id(pci_conf, PCI_DEVICE_ID_INTEL_82371SB_2);
return usb_uhci_common_initfn(s);
}
static int usb_uhci_piix4_initfn(PCIDevice *dev)
{
UHCIState *s = DO_UPCAST(UHCIState, dev, dev);
uint8_t *pci_conf = s->dev.config;
pci_config_set_vendor_id(pci_conf, PCI_VENDOR_ID_INTEL);
pci_config_set_device_id(pci_conf, PCI_DEVICE_ID_INTEL_82371AB_2);
return usb_uhci_common_initfn(s);
}
static int usb_uhci_vt82c686b_initfn(PCIDevice *dev)
{
UHCIState *s = DO_UPCAST(UHCIState, dev, dev);
uint8_t *pci_conf = s->dev.config;
pci_config_set_vendor_id(pci_conf, PCI_VENDOR_ID_VIA);
pci_config_set_device_id(pci_conf, PCI_DEVICE_ID_VIA_UHCI);
/* USB misc control 1/2 */
pci_set_long(pci_conf + 0x40,0x00001000);
/* PM capability */
pci_set_long(pci_conf + 0x80,0x00020001);
/* USB legacy support */
pci_set_long(pci_conf + 0xc0,0x00002000);
return usb_uhci_common_initfn(s);
}
static PCIDeviceInfo uhci_info[] = {
{
.qdev.name = "piix3-usb-uhci",
.qdev.size = sizeof(UHCIState),
.qdev.vmsd = &vmstate_uhci,
.init = usb_uhci_piix3_initfn,
},{
.qdev.name = "piix4-usb-uhci",
.qdev.size = sizeof(UHCIState),
.qdev.vmsd = &vmstate_uhci,
.init = usb_uhci_piix4_initfn,
},{
.qdev.name = "vt82c686b-usb-uhci",
.qdev.size = sizeof(UHCIState),
.qdev.vmsd = &vmstate_uhci,
.init = usb_uhci_vt82c686b_initfn,
},{
/* end of list */
}
};
static void uhci_register(void)
{
pci_qdev_register_many(uhci_info);
}
device_init(uhci_register);
void usb_uhci_piix3_init(PCIBus *bus, int devfn)
{
pci_create_simple(bus, devfn, "piix3-usb-uhci");
}
uhci: rewrite UHCI emulator, fully async operation with multiple outstanding transactions (Max Krasnyansky) This is esentially a re-write of the QEMU UHCI layer. My initial goal was to support fully async operation with multiple outstanding async transactions. Along the way I realized that I can greatly simplify and cleanup the overall logic. There was a lot of duplicate and confusing code in the UHCI data structure parsing and other places. We were actually violating UHCI spec in handling async ISOC transaction (host controller is not supposed to write into the frame pointer). The reason I wanted to support fully async operation is because current synchronous version is unusable with most devices exported from host (via usb-linux.c). Transactions take a long time and the whole VM becomes slow as hell. Current async support is very rudimentory and for the most part non-functional. Single transaction at a time is simply not enough. I have a device for which XP driver submits both IN and OUT packets at the same time. IN packet always times out unless OUT packet makes it to the device. Hence we must be able to process both in order for that device to work. The new code is backwards compatible and was first tested agains original synchronous usb-linux.c and builtin usb devices like tablet which is also synchronous. Rewrite of the usb-linux.c is coming up next. Async support was tested against various XP versions (ie XP, SP2, SP3) and a bunch of different USB devices: serial port controllers, mice, keyboard, JTAG dongles (from Xilinx and Altera). ISOC support was only lighly tested and needs more work. It's not any worse than current code though. UHCI parser changes are probably somewhat hard to review without the understanding of the UHCI spec. The async design should be fairly easy to follow. Basically we have a list of async objects for each pending transfer. Async objects are tagged with the original TD (transfer descriptor) address and token. We now support unlimited number of outstanding isoc and one outstanding bulk/intr/ctrl transfer per QH (queue head). UHCI spec does not have a clear protocol for the cancelation of the trasfer requests. Driver can yank out TDs on any frame boundary. In oder to handle that I added somewhat fancy TD validation logic logic to avoid unnecessary cancelations. Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5050 c046a42c-6fe2-441c-8c8c-71466251a162
2008-08-21 23:30:31 +04:00
void usb_uhci_piix4_init(PCIBus *bus, int devfn)
{
pci_create_simple(bus, devfn, "piix4-usb-uhci");
}
void usb_uhci_vt82c686b_init(PCIBus *bus, int devfn)
{
pci_create_simple(bus, devfn, "vt82c686b-usb-uhci");
}