calling scsipi_make_xs() and scsipi_execute_xs(), instead push these into
scsipi_command. Make bustype_cmd and PHOLD/PRELE be called from
scsipi_execute_xs(). This allows us to create a xfer structure -- possibly on
the stack -- and call scsipi_execute_xs() directly.
scsipi_done() from their scsipi_request().
For this, add a struct scsipi_xfer * argument to scsipi_command().
If not NULL scsipi_command() will use this to enqueue this xfer, otherwise
it'll try to allocate a new one. This scsipi_xfer has to be allocated
and initialised by scsipi_make_xs() or equivalent.
In xxstart(), allocate a scsipi_xfer using scsipi_make_xs(), and if not NULL,
dequeue the buffer before calling scsipi_command(). This makes sure that
scsipi_command() will not fail, and also makes sure that xxstart() won't
be called again between the BUFQ_PEEK() and BUFQ_GET().
Fix "dequeued wrong buf" panics reported by Juergen Hannken-Illjes in
private mail and Andreas Wrede on current-users@.
Thanks to Jason Thorpe and Chuck Silver for review, and Andreas Wrede for
testing the patch.
sd3(mpt0:0:1:0): unable to allocate scsipi_xfer
sd3: not queued, error 12
Havard Eidnes's analysis of this problem is that the scsipi_xfer pool is
competing for resources with other pools, including the the inode and vnode
pools which can grow quite large.
*_scsipi_cmd(): don't biodone the buffer if scsipi_make_xs() fails, let the
caller deal with the problem
start function of block devices drivers: dequeue the buffer after the
scsipi_command() call. If scsipi_command() fails with ENOMEM don't dequeue
the buffer, and schedule a callout to call the start function after
some delay.
scsipi_init(): prime the scsipi_xfer_pool with one page. This ensure that
there is always some scsipi_xfer to play with. If scsipi_command() fails
because of pool_get(), we're sure there will be resources available later,
when the pending commands have completed.
Reviewed by Jason Thorpe and Havard Eidnes.
Todo: remove the "unable to allocate scsipi_xfer" and "not queued, error %d"
printfs, but I choose to keep them for now, to help make sure the code does
what it should.
The controller is handling them, calling scsipi_done() here will end up in the
xfer being scsipi_done()'ed a second time when it completes in the controller
code. In addition, the way the loop was done here would end up in an infinite
loop, because the channel kernel thread needs to run to remove a command from
this queue.
scsibusdetach(): scsipi_done() all commands from periph_xferq. The controller
is already gone, and these commands will never complete.
Shut down the channel (which will cause the kenrel thread to exit) after
detaching the childs, as they will need the kernel thread for
scsipi_wait_drain().
Fix kernel hang or deadlock when detaching devices (either by scsictl detach
or unplug) with active commands.
This is a completely rewritten scsipi_xfer execution engine, and the
associated changes to HBA drivers. Overview of changes & features:
- All xfers are queued in the mid-layer, rather than doing so in an
ad-hoc fashion in individual adapter drivers.
- Adapter/channel resource management in the mid-layer, avoids even trying
to start running an xfer if the adapter/channel doesn't have the resources.
- Better communication between the mid-layer and the adapters.
- Asynchronous event notification mechanism from adapter to mid-layer and
peripherals.
- Better peripheral queue management: freeze/thaw, sorted requeueing during
recovery, etc.
- Clean separation of peripherals, adapters, and adapter channels (no more
scsipi_link).
- Kernel thread for each scsipi_channel makes error recovery much easier
(no more dealing with interrupt context when recovering from an error).
- Mid-layer support for tagged queueing: commands can have the tag type
set explicitly, tag IDs are allocated in the mid-layer (thus eliminating
the need to use buggy tag ID allocation schemes in many adapter drivers).
- support for QUEUE FULL and CHECK CONDITION status in mid-layer; the command
will be requeued, or a REQUEST SENSE will be sent as appropriate.
Just before the merge syssrc has been tagged with thorpej_scsipi_beforemerge
to remove all xfers from the pending queue. It removes only xfers for
asynchronous transactions. So, simply loop over all pending xfers
with calling scsipi_done and wait xfers to drain. Addresses PR#9703.
- `flags' is now gone, replaced with `xs_control' and `xs_status'.
- Massive cleanup of the control flags. Now we explicitly say that
a job is to complete asynchronously, rather than relying on side-effects,
and use a new flag to now that device discovery is being performed.
- Do SCSI device discovery interrupt-driven.
scsi_base.c to scsipi_base.c. Rename the functions from scsi_verbose.c
too, and rename the file itself. Cleaup includes too (scsi_*.h should not
be #included in scsipi_*.h files, which are supposed to be
common to atapi and scsi).
UVM was written by chuck cranor <chuck@maria.wustl.edu>, with some
minor portions derived from the old Mach code. i provided some help
getting swap and paging working, and other bug fixes/ideas. chuck
silvers <chuq@chuq.com> also provided some other fixes.
this is the rest of the MI portion changes.
this will be KNF'd shortly. :-)
fix 'cd' driver's NCD_SCSI bogosity (was using testing wrong macro!)
clean up in various ways:
* make common atapi_mode_{sense,select}() functions.
* put ATAPI data structures in more sensible headers, split up by
device type.
* include headers a bit more carefully.
* pass flags to attachment-specific cd functions, and use them.
* get rid of SCSI bits in scsipi_base.h's scsipi_make_xs(), move
them into the correct place in scsi_base.c.
* fix minor typo in struct name in scsipiconf.h (which was apparently
never used except in a #define later in the same file).
* use __attribute__ to force 4-byte alignment for xs command store,
so that architectures trying to bus_space_write_multi_N() (where
N > 1) that data to a controller won't lose.
* clean up a few comments in typos, and make a few #defines easier to
understand/maintain.
* rename cd_link.h to cdvar.h (via repository copy). This is exactly
what a 'var' file is supposed to be.
- Indent with tab of width 8.
- Use four column to indent continuation line.
- Fold long line if possible.
- Use return (xx) instead of return xx.
- Compare pointer against NULL instead of testing like boolean.
- Delete whitespace at the end of line.
- Delete whitespace in front of function call operator.
- Delete whitespace after cast.
- Dereference a pointer to function explicitly.
- Add an empty line after local variable declaration.
- Use NULL instead of (char *)0.
- Dont use block for single statement.
(currently only CD-ROM drives on i386). The sys/dev/scsipi system provides 2
busses to which devices can attach (scsibus and atapibus). This needed to
change some include files and structure names in the low level scsi drivers.
gup to the kernel) more detailed SCSI error printing, based off of final
SCSI-2 specification. This will also print out things like progress indicator
errors, and will also dump out additional sense bytes that haven't been decoded.
Also, turned the error message 'DELAYED' to 'DEFERRED' since that's more
accurate.
is called *after* the driver `done' routine. This fixes disk I/O statistics
on SCSI devices.
Also, calling the `done' routine with a `complete' argument of 0 and actually
having it do anything meaningful loses in at least 3 ways, so just nuke the
argument altogether and don't call it this way. If the driver needs to do
some error handling, that's what `err_handler' is for.
into 8 bits generate (bogus) warnings on some architectures, but the
change to pad on "scsi inquiry" is no longer needed, apparently.
Thanks to Matthias Pfaller for pointing the latter out to me.
- Deal with devices that ignore the length specified in the
inquiry command.
- Allow asynchronous requests without using a buf (key off NOSLEEP instead).
if the version is <= SCSI-2. This should help some older SCSI
devices that previously needed the "NOLUNS" quirk. While this is
not strictly necessary on SCSI-2 devices, the spec allows it,
so we set it for SCSI-2 devices "just in case". See section 7.2.2 of
Draft X3T9.2 Rev 10L for details.
christos