Commit Graph

16 Commits

Author SHA1 Message Date
thorpej
db36913c87 The code that creates/destroys SGMAP DMA maps is the same; put it
in a common place and share it.
2001-01-03 19:15:59 +00:00
mrg
c88e94a407 remove include of <vm/vm.h>. <vm/vm.h> -> <uvm/uvm_extern.h> 2000-06-29 08:58:45 +00:00
thorpej
97eba73a40 If we have a Pyxis with the DMA page crossing bug, don't allow coalescing
of adjacent DMA segments.

XXX This is still not perfect... but making it perfect will probably
require additions to the bus_dma interface and the ISA autoconfiguration
interface.
2000-02-06 01:26:50 +00:00
thorpej
35f3518a91 Fix a fatal typo in a Pyxis SGMAP TLB bug workaround. Noticed by
Jeff Roberson <nomad@nop.aliensystems.com>.
2000-01-25 03:32:36 +00:00
thorpej
d5df55112a vm_offset_t -> {paddr_t,vaddr_t}, vm_size_t -> vsize_t 1998-08-14 16:50:00 +00:00
thorpej
e2ebc10c2d Duuuh! Align the SGMAP page tables to 32K, not 32M. 1998-06-23 02:31:05 +00:00
thorpej
c0fa3c6ac4 Don't call *_dma_init() twice; there's no need to. Just do it in *attach(). 1998-06-06 01:33:23 +00:00
thorpej
616125f8d1 Deal with a hardware bug in Pass 1 and Pass 2 Pyxis chips. Basically,
the scatter/gather TLB cannot be invalidated on these chips.  So, to
work around this, we configure the otherwise unsed DMA Window 2 as a
2M SGMAP window at 128M, point all of its page table entries at the
DMA spill page, and, when the TLB is to be invalidated, put the PCI bus
into loopback mode, and create a target hit on Window 2 every 64k for
the number of TLB entries (plus a few ... it seems to not work unless
we read a few extra times), forcing out old TLB entries to make room for
the new, dummy target hits.
1998-06-04 18:11:23 +00:00
thorpej
d6041754c2 Allow the DMA tag to specify a boundary contraint. If the device has a
more strict boundary, the map will use it, otherwise the map will inherit
the tag's, unless the tag's constraint is 0 (no boundary constraint).
1998-06-03 18:25:53 +00:00
thorpej
49aa171ab9 Add support for chaining DMA windows together, for falling back on
SGMAPs if a direct-mapped window fails.
1998-05-13 21:21:16 +00:00
thorpej
4665241bf8 Simplify the direct-mapped DMA case somewhat by adding a window base
member to the DMA tag, and calling the direct-mapped back-ends directly,
rather than through chipset-specific front-ends which pass the window
base as an additional argument.
1998-05-07 20:09:37 +00:00
thorpej
612aea038c Use the common _bus_dmamap_sync() as the _dmamap_sync method in the
bus_dma_tag_t.
1998-02-04 07:37:28 +00:00
thorpej
3366b987d0 Put SGMAP-related stuff in the DMA map structure directly, rather than
indirecting through a pointer.
1998-01-17 21:53:52 +00:00
thorpej
63db8935a6 Fix initialization of DMA window 0:
- Make sure the page table is aligned to at least 32k.
- Don't consider the MEMCS signal when checking for a window hit.
1998-01-17 03:43:59 +00:00
thorpej
693b3e7262 Nuke the idea of <machine/options.h>. It completely defeats the purpose
of fine-grain option dependencies.
1997-09-02 13:17:47 +00:00
thorpej
46b89d7739 Pull thorpej-bus-dma branch into mainline. 1997-06-06 23:54:24 +00:00