1. Implement a large set of ioctl shims so that the Linux management apps
from LSI will work. This includes infrastructure to support adding, deleting
and rescanning arrays at runtime. This is based on work from Doug Ambrosko,
heavily augmented by LSI and Yahoo.
2. Implement full 64-bit DMA support. Systems with more than 4GB of RAM
can now operate without the cost of bounce buffers. Cards that cannot do
64-bit DMA will automatically revert to using bounce buffers. This option
can be forced off by setting the 'hw.amr.force_sg32" tunable in the loader.
It should only be turned off for debugging purposes. This work was sponsored
by Yahoo.
3. Streamline the command delivery and interrupt handler paths after
much discussion with Dell and LSI. The logic now closely matches the
intended design, making it both more robust and much faster. Certain
i/o failures under heavy load should be fixed with this.
4. Optimize the locking. In the interrupt handler, the card can be checked
for completed commands without any locks held, due to the handler being
implicitely serialized and there being no need to look at any shared data.
Only grab the lock to return the command structure to the free pool. A
small optimization can still be made to collect all of the completions
together and then free them together under a single lock.
Items 3 and 4 significantly increase the performance of the driver. On an
LSI 320-2X card, transactions per second went from 13,000 to 31,000 in my
testing with these changes. However, these changes are still fairly
experimental and shouldn't be merged to 6.x until there is more testing.
Thanks to Doug Ambrosko, LSI, Dell, and Yahoo for contributing towards
this.
- Introduce the amr_io_lock to control access to command queues, bio queues,
and the hardware.
- Eliminate the taskqueue and do all completion processing in the ithread.
- Assign a static slot number to each command instead of doing a linear
search for free slots each time a command is needed.
- Modify the interrupt handler to more closely match what Linux does, for
safety.
Previously the "struct disk" were owned by the device driver and this
gave us problems when the device disappared and the users of that device
were not immediately disappearing.
Now the struct disk is allocate with a new call, disk_alloc() and owned
by geom_disk and just abandonned by the device driver when disk_create()
is called.
Unfortunately, this results in a ton of "s/\./->/" changes to device
drivers.
Since I'm doing the sweep anyway, a couple of other API improvements
have been carried out at the same time:
The Giant awareness flag has been flipped from DISKFLAG_NOGIANT to
DISKFLAG_NEEDSGIANT
A version number have been added to disk_create() so that we can detect,
report and ignore binary drivers with old ABI in the future.
Manual page update to follow shortly.
Retain the mistake of not updating the devstat API for now.
Spell bioq_disksort() consistently with the remaining bioq_*().
#include <geom/geom_disk.h> where this is more appropriate.
in geom_disk.c.
As a side effect this makes a lot of #include <sys/devicestat.h>
lines not needed and some biofinish() calls can be reduced to
biodone() again.
Change the argument to disk_destroy() to be the same struct disk * as
disk_create() takes.
This enables drivers to ignore the (now) bogus dev_t which disk_create()
returns.
as module. This also fix's issue kern/45713.
Fix - polling was implemented incorrectly for
adapter enquiry and adapter flush.
(2) Problem: PANIC when unloading driver
as module.
Fix - device nodes are not destroyed for amr0,
and amrd* when driver is unloaded
(3) Problem: PANIC from loading driver when
3ware adapter present, error message "Warning
"amrd is usurping twed's bmaj"
Fix - put #idef freebsd version < 500000 for
bmaj -1 -> amrd_cdevsw
(4) Problem: warnings in driver when compiling
with DAMR_DEBUG param enabled in Makefile
Fix - fix the warnings so driver can compile
when -Werror is present in Makefile.
Approved by: jhb
MFC: 7 days
amr_cam.c,
amrreg.h,
amrvar.h:
- added support for 12/16 byte cdb's, effecting CAM branch only ( non-disk support )
amrreg.h:
- increased number of scatter gather elements from 16 to 26.
amr_pci.c:
- amr_pci_free(), incorrect bus tag meant for 'amr_mailbox_dmat' was being freed
all:
- copyright change requested by scottl
Reviewed by: ps,scottl
MFC after: 1 week
changes for "LSILogic"
(2) enabled non-disk support through CAM interface
(3) HA_INQ (a) enabled tagged queuing (b) disable reset during
driver loading (b) renamed BSDi string to LSI
(4) disabled detecting disk devices during SCSI INQUIRY
(5) changed dcdb single element sglist to send one entire buffer chunk
(6) nsgelem not set in sglist
(7) ap_data_transfer_length not set for dcdb
(8) changed "struct thread" to "d_thread_t" for compatibliity { xxx_open,
xxx_close, xxx_ioctl }
(9) miscellaneous compatiblity fixes
(10) bug fix for 0x0409/0x1000 card
(11) added compiling amr_cam.c in sys/conf/files
(12) added compiling amr_cam.c in sys/modules/amr/Makefile
Reviewed by:ps
MFC after:1 week
1 week
and predictable way, and I apologize if I have gotten it wrong anywhere,
getting prior review on a patch like this is not feasible, considering
the number of people involved and hardware availability etc.)
If struct disklabel is the messenger: kill the messenger.
Inside struct disk we had a struct disklabel which disk drivers used to
communicate certain metrics to the disklayer above (GEOM or the disk
mini-layer). This commit changes this communication to use four
explicit fields instead.
Amongst the benefits is that the fields do not get overwritten by
wrong or bogus on-disk disklabels.
Once that is clear, <sys/disk.h> which is included in the drivers
no longer need to pull <sys/disklabel.h> and <sys/diskslice.h> in,
the few places that needs them, have gotten explicit #includes for
them.
The disklabel inside struct disk is now only for internal use in
the disk mini-layer, so instead of embedding it, we malloc it as
we need it.
This concludes (modulus any mistakes) the series of disklabel related
commits.
I belive it all amounts to a NOP for all the rest of you :-)
Sponsored by: DARPA & NAI Labs.
Note ALL MODULES MUST BE RECOMPILED
make the kernel aware that there are smaller units of scheduling than the
process. (but only allow one thread per process at this time).
This is functionally equivalent to teh previousl -current except
that there is a thread associated with each process.
Sorry john! (your next MFC will be a doosie!)
Reviewed by: peter@freebsd.org, dillon@freebsd.org
X-MFC after: ha ha ha ha
- New support for 40LD firmware found in Series 475 and 471 adapters.
- Better support for 8LD firmware adapters
- Ioctl passthrough interface for userland utilities.
- Improved error handling and queueing.
- Several bugfixes (including the 'still open' shutdown bug and
closing some small race conditions).
- Zone-style command allocator, reducing memory wasted under heavy
load conditions.
- CAM interface (disabled and not fully working) for SCSI passthrough
access to non-disk devices
Thanks to AMI for supplying a pile of new adapters and various other
help in making this happen.
<sys/bio.h>.
<sys/bio.h> is now a prerequisite for <sys/buf.h> but it shall
not be made a nested include according to bdes teachings on the
subject of nested includes.
Diskdrivers and similar stuff below specfs::strategy() should no
longer need to include <sys/buf.> unless they need caching of data.
Still a few bogus uses of struct buf to track down.
Repocopy by: peter
Exceptions:
Vinum untouched. This means that it cannot be compiled.
Greg Lehey is on the case.
CCD not converted yet, casts to struct buf (still safe)
atapi-cd casts to struct buf to examine B_PHYS
(Much of this done by script)
Move B_ORDERED flag to b_ioflags and call it BIO_ORDERED.
Move b_pblkno and b_iodone_chain to struct bio while we transition, they
will be obsoleted once bio structs chain/stack.
Add bio_queue field for struct bio aware disksort.
Address a lot of stylistic issues brought up by bde.
- Add periodic status monitoring routine. Currently just detects
lost commands, further functionality pending data from AMI.
Add some new commands states; WEDGED (never coming back) and
LATE (for when a command that wasmarked as WEDGED comes bacj,
- Remove a number of redundant efforts to poll the card for completed
commands. This is what interrupt handlers are for.
- Limit the maximum number of outstanding I/O transactions. It seems
that some controllers report more than they can really handle,
and exceding this limit can cause the controller to lock up.
- Don't use 'wait' mode for anything where the controller might not
be able to generate interrupts. (Keep the 'wait' mode though sa it
will become useful when we start taking userspace commands.
- Use a similar atomic locking trategy to the Mylex driver to prevent
some reentrancy problems.
- Correctly calculate the block count for non-whoile-bloch transfers
(actually illegal).
- Use the dsik device's si_drv1 field instead of b_driver1 in the
buf struct to pass the driver identifier arond.
- Rewrite amr_start and amr_done() along the lines of the Mylex driver
in order to improve robustnes.
- Always force the PCI busmaster bit on.
The same goes for CD drivers and tape drivers. In systems with mixed IDE
and SCSI, devices in the same priority class will be sorted in attach
order.
Also, the 'CCD' priority is now the 'ARRAY' priority, and a number of
drivers have been modified to use that priority.
This includes the necessary changes to all drivers, except the ATA drivers.
Soren will modify those separately.
This does not include and does not require any change in the devstat
version number, since no known userland applications use the priority
enumerations.
Reviewed by: msmith, sos, phk, jlemon, mjacob, bde
the time spent at splbio(). We now avoid it unless we are actually
manipulating the command queues themselves. This doesn't improve
performance noticeably, but should improve concurrency somewhat.
the AMI PCI controllers using the 8LD firmware interface (40LD firmware
will be supported as soon as I have hardware to test with).
These controllers are rebadged by Dell as the PERC, as well as by HP
and possibly other vendors.
