Damien Le Moal [Thu, 14 May 2020 05:56:26 +0000 (14:56 +0900)]
nvme: fix io_opt limit setting
Currently, a namespace io_opt queue limit is set by default to the
physical sector size of the namespace and to the the write optimal
size (NOWS) when the namespace reports optimal IO sizes. This causes
problems with block limits stacking in blk_stack_limits() when a
namespace block device is combined with an HDD which generally do not
report any optimal transfer size (io_opt limit is 0). The code:
/* Optimal I/O a multiple of the physical block size? */
if (t->io_opt & (t->physical_block_size - 1)) {
t->io_opt = 0;
t->misaligned = 1;
ret = -1;
}
in blk_stack_limits() results in an error return for this function when
the combined devices have different but compatible physical sector
sizes (e.g. 512B sector SSD with 4KB sector disks).
Fix this by not setting the optimal IO size queue limit if the namespace
does not report an optimal write size value.
Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Bart van Assche <bvanassche@acm.org> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Christoph Hellwig <hch@lst.de>
Martin George [Tue, 12 May 2020 16:47:04 +0000 (22:17 +0530)]
nvme-fc: print proper nvme-fc devloss_tmo value
The nvme-fc devloss_tmo is computed as the min of either the
ctrl_loss_tmo (max_retries * reconnect_delay) or the remote port's
devloss_tmo. But what gets printed as the nvme-fc devloss_tmo in
nvme_fc_reconnect_or_delete() is always the remote port's devloss_tmo
value. So correct this by printing the min value instead.
Signed-off-by: Martin George <marting@netapp.com> Reviewed-by: James Smart <james.smart@broadcom.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
Sagi Grimberg [Mon, 18 May 2020 17:47:48 +0000 (10:47 -0700)]
nvmet-tcp: move send/recv error handling in the send/recv methods instead of call-sites
Have routines handle errors and just bail out of the poll loop.
This simplifies the code and will help as we may enhance the poll
loop logic and these are somewhat in the way.
Signed-off-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de>
Jiri Kosina [Tue, 26 May 2020 09:49:18 +0000 (11:49 +0200)]
block/floppy: fix contended case in floppy_queue_rq()
Since the switch of floppy driver to blk-mq, the contended (fdc_busy) case
in floppy_queue_rq() is not handled correctly.
In case we reach floppy_queue_rq() with fdc_busy set (i.e. with the floppy
locked due to another request still being in-flight), we put the request
on the list of requests and return BLK_STS_OK to the block core, without
actually scheduling delayed work / doing further processing of the
request. This means that processing of this request is postponed until
another request comes and passess uncontended.
Which in some cases might actually never happen and we keep waiting
indefinitely. The simple testcase is
for i in `seq 1 2000`; do echo -en $i '\r'; blkid --info /dev/fd0 2> /dev/null; done
run in quemu. That reliably causes blkid eventually indefinitely hanging
in __floppy_read_block_0() waiting for completion, as the BIO callback
never happens, and no further IO is ever submitted on the (non-existent)
floppy device. This was observed reliably on qemu-emulated device.
Fix that by not queuing the request in the contended case, and return
BLK_STS_RESOURCE instead, so that blk core handles the request
rescheduling and let it pass properly non-contended later.
Stefan Haberland [Tue, 19 May 2020 14:22:59 +0000 (16:22 +0200)]
s390/dasd: remove ioctl_by_bdev calls
The IBM partition parser requires device type specific information only
available to the DASD driver to correctly register partitions. The
current approach of using ioctl_by_bdev with a fake user space pointer
is discouraged.
Fix this by replacing IOCTL calls with direct in-kernel function calls.
Suggested-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Stefan Haberland <sth@linux.ibm.com> Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com> Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Martijn Coenen [Wed, 13 May 2020 13:38:45 +0000 (15:38 +0200)]
loop: Add LOOP_CONFIGURE ioctl
This allows userspace to completely setup a loop device with a single
ioctl, removing the in-between state where the device can be partially
configured - eg the loop device has a backing file associated with it,
but is reading from the wrong offset.
Besides removing the intermediate state, another big benefit of this
ioctl is that LOOP_SET_STATUS can be slow; the main reason for this
slowness is that LOOP_SET_STATUS(64) calls blk_mq_freeze_queue() to
freeze the associated queue; this requires waiting for RCU
synchronization, which I've measured can take about 15-20ms on this
device on average.
In addition to doing what LOOP_SET_STATUS can do, LOOP_CONFIGURE can
also be used to:
- Set the correct block size immediately by setting
loop_config.block_size (avoids LOOP_SET_BLOCK_SIZE)
- Explicitly request direct I/O mode by setting LO_FLAGS_DIRECT_IO
in loop_config.info.lo_flags (avoids LOOP_SET_DIRECT_IO)
- Explicitly request read-only mode by setting LO_FLAGS_READ_ONLY
in loop_config.info.lo_flags
Here's setting up ~70 regular loop devices with an offset on an x86
Android device, using LOOP_SET_FD and LOOP_SET_STATUS:
vsoc_x86:/system/apex # time for i in `seq 30 100`;
do losetup -r -o 4096 /dev/block/loop$i com.android.adbd.apex; done
0m03.40s real 0m00.02s user 0m00.03s system
Here's configuring ~70 devices in the same way, but using a modified
losetup that uses the new LOOP_CONFIGURE ioctl:
vsoc_x86:/system/apex # time for i in `seq 30 100`;
do losetup -r -o 4096 /dev/block/loop$i com.android.adbd.apex; done
0m01.94s real 0m00.01s user 0m00.01s system
Martijn Coenen [Wed, 13 May 2020 13:38:44 +0000 (15:38 +0200)]
loop: Clean up LOOP_SET_STATUS lo_flags handling
LOOP_SET_STATUS(64) will actually allow some lo_flags to be modified; in
particular, LO_FLAGS_AUTOCLEAR can be set and cleared, whereas
LO_FLAGS_PARTSCAN can be set to request a partition scan. Make this
explicit by updating the UAPI to include the flags that can be
set/cleared using this ioctl.
The implementation can then blindly take over the passed in flags,
and use the previous flags for those flags that can't be set / cleared
using LOOP_SET_STATUS.
Martijn Coenen [Wed, 13 May 2020 13:38:42 +0000 (15:38 +0200)]
loop: Move loop_set_status_from_info() and friends up
So we can use it without forward declaration. This is a separate commit
to make it easier to verify that this is just a move, without functional
modifications.
Martijn Coenen [Wed, 13 May 2020 13:38:39 +0000 (15:38 +0200)]
loop: Refactor loop_set_status() size calculation
figure_loop_size() calculates the loop size based on the passed in
parameters, but at the same time it updates the offset and sizelimit
parameters in the loop device configuration. That is a somewhat
unexpected side effect of a function with this name, and it is only only
needed by one of the two callers of this function - loop_set_status().
Move the lo_offset and lo_sizelimit assignment back into loop_set_status(),
and use the newly factored out functions to validate and apply the newly
calculated size. This allows us to get rid of figure_loop_size() in a
follow-up commit.
Martijn Coenen [Wed, 13 May 2020 13:38:35 +0000 (15:38 +0200)]
loop: Call loop_config_discard() only after new config is applied
loop_set_status() calls loop_config_discard() to configure discard for
the loop device; however, the discard configuration depends on whether
the loop device uses encryption, and when we call it the encryption
configuration has not been updated yet. Move the call down so we apply
the correct discard configuration based on the new configuration.
Signed-off-by: Martijn Coenen <maco@android.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Bob Liu <bob.liu@oracle.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Jens Axboe [Wed, 13 May 2020 21:17:01 +0000 (15:17 -0600)]
Merge branch 'md-next' of git://git.kernel.org/pub/scm/linux/kernel/git/song/md into for-5.8/drivers
Pull MD changes from Song.
* 'md-next' of git://git.kernel.org/pub/scm/linux/kernel/git/song/md:
md/raid1: Replace zero-length array with flexible-array
md: add a newline when printing parameter 'start_ro' by sysfs
md: stop using ->queuedata
md/raid1: release pending accounting for an I/O only after write-behind is also finished
md: remove redundant memalloc scope API usage
raid5: update code comment of scribble_alloc()
raid5: remove gfp flags from scribble_alloc()
md: use memalloc scope APIs in mddev_suspend()/mddev_resume()
md: remove the extra line for ->hot_add_disk
md: flush md_rdev_misc_wq for HOT_ADD_DISK case
md: don't flush workqueue unconditionally in md_open
md: add new workqueue for delete rdev
md: add checkings before flush md_misc_wq
md/raid1: Replace zero-length array with flexible-array
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
sizeof(flexible-array-member) triggers a warning because flexible array
members have incomplete type[1]. There are some instances of code in
which the sizeof operator is being incorrectly/erroneously applied to
zero-length arrays and the result is zero. Such instances may be hiding
some bugs. So, this work (flexible-array member conversions) will also
help to get completely rid of those sorts of issues.
David Jeffery [Mon, 27 Jan 2020 15:26:19 +0000 (10:26 -0500)]
md/raid1: release pending accounting for an I/O only after write-behind is also finished
When using RAID1 and write-behind, md can deadlock when errors occur. With
write-behind, r1bio structs can be accounted by raid1 as queued but not
counted as pending. The pending count is dropped when the original bio is
returned complete but write-behind for the r1bio may still be active.
This breaks the accounting used in some conditions to know when the raid1
md device has reached an idle state. It can result in calls to
freeze_array deadlocking. freeze_array will never complete from a negative
"unqueued" value being calculated due to a queued count larger than the
pending count.
To properly account for write-behind, move the call to allow_barrier from
call_bio_endio to raid_end_bio_io. When using write-behind, md can call
call_bio_endio before all write-behind I/O is complete. Using
raid_end_bio_io for the point to call allow_barrier will release the
pending count at a point where all I/O for an r1bio, even write-behind, is
done.
Signed-off-by: David Jeffery <djeffery@redhat.com> Signed-off-by: Song Liu <songliubraving@fb.com>
Coly Li [Thu, 9 Apr 2020 14:17:23 +0000 (22:17 +0800)]
md: remove redundant memalloc scope API usage
In mddev_create_serial_pool(), memalloc scope APIs memalloc_noio_save()
and memalloc_noio_restore() are used when allocating memory by calling
mempool_create_kmalloc_pool(). After adding the memalloc scope APIs in
raid array suspend context, it is unncessary to explicitly call them
around mempool_create_kmalloc_pool() any longer.
This patch removes the redundant memalloc scope APIs in
mddev_create_serial_pool().
Signed-off-by: Coly Li <colyli@suse.de> Cc: Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by: Song Liu <songliubraving@fb.com>
Coly Li [Thu, 9 Apr 2020 14:17:21 +0000 (22:17 +0800)]
raid5: remove gfp flags from scribble_alloc()
Using GFP_NOIO flag to call scribble_alloc() from resize_chunk() does
not have the expected behavior. kvmalloc_array() inside scribble_alloc()
which receives the GFP_NOIO flag will eventually call kmalloc_node() to
allocate physically continuous pages.
Now we have memalloc scope APIs in mddev_suspend()/mddev_resume() to
prevent memory reclaim I/Os during raid array suspend context, calling
to kvmalloc_array() with GFP_KERNEL flag may avoid deadlock of recursive
I/O as expected.
This patch removes the useless gfp flags from parameters list of
scribble_alloc(), and call kvmalloc_array() with GFP_KERNEL flag. The
incorrect GFP_NOIO flag does not exist anymore.
Fixes: b330e6a49dc3 ("md: convert to kvmalloc") Suggested-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Song Liu <songliubraving@fb.com>
Coly Li [Thu, 9 Apr 2020 14:17:20 +0000 (22:17 +0800)]
md: use memalloc scope APIs in mddev_suspend()/mddev_resume()
In raid5.c:resize_chunk(), scribble_alloc() is called with GFP_NOIO
flag, then it is sent into kvmalloc_array() inside scribble_alloc().
The problem is kvmalloc_array() eventually calls kvmalloc_node() which
does not accept non GFP_KERNEL compatible flag like GFP_NOIO, then
kmalloc_node() is called indeed to allocate physically continuous
pages. When system memory is under heavy pressure, and the requesting
size is large, there is high probability that allocating continueous
pages will fail.
But simply using GFP_KERNEL flag to call kvmalloc_array() is also
progblematic. In the code path where scribble_alloc() is called, the
raid array is suspended, if kvmalloc_node() triggers memory reclaim I/Os
and such I/Os go back to the suspend raid array, deadlock will happen.
What is desired here is to allocate non-physically (a.k.a virtually)
continuous pages and avoid memory reclaim I/Os. Michal Hocko suggests
to use the mmealloc sceope APIs to restrict memory reclaim I/O in
allocating context, specifically to call memalloc_noio_save() when
suspend the raid array and to call memalloc_noio_restore() when
resume the raid array.
This patch adds the memalloc scope APIs in mddev_suspend() and
mddev_resume(), to restrict memory reclaim I/Os during the raid array
is suspended. The benifit of adding the memalloc scope API in the
unified entry point mddev_suspend()/mddev_resume() is, no matter which
md raid array type (personality), we are sure the deadlock by recursive
memory reclaim I/O won't happen on the suspending context.
Please notice that the memalloc scope APIs only take effect on the raid
array suspending context, if the memory allocation is from another new
created kthread after raid array suspended, the recursive memory reclaim
I/Os won't be restricted. The mddev_suspend()/mddev_resume() entries are
used for the critical section where the raid metadata is modifying,
creating a kthread to allocate memory inside the critical section is
queer and very probably being buggy.
Fixes: b330e6a49dc3 ("md: convert to kvmalloc") Suggested-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Coly Li <colyli@suse.de> Signed-off-by: Song Liu <songliubraving@fb.com>
It is not not necessary to add a newline for them since they don't exceed
80 characters, and it is not intutive to distinguish ->hot_add_disk() from
hot_add_disk() too.
Signed-off-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by: Song Liu <songliubraving@fb.com>
Since rdev->kobj is removed asynchronously, it is possible that the
rdev->kobj still exists when try to add the rdev again after rdev
is removed. But this path md_ioctl (HOT_ADD_DISK) -> hot_add_disk
-> bind_rdev_to_array missed it.
Signed-off-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by: Song Liu <songliubraving@fb.com>
md: don't flush workqueue unconditionally in md_open
We need to check mddev->del_work before flush workqueu since the purpose
of flush is to ensure the previous md is disappeared. Otherwise the similar
deadlock appeared if LOCKDEP is enabled, it is due to md_open holds the
bdev->bd_mutex before flush workqueue.
And md_alloc also flushed the same workqueue, but the thing is different
here. Because all the paths call md_alloc don't hold bdev->bd_mutex, and
the flush is necessary to avoid race condition, so leave it as it is.
Signed-off-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by: Song Liu <songliubraving@fb.com>
Since the purpose of call flush_workqueue in new_dev_store is to ensure
md_delayed_delete() has completed, so we should check rdev->del_work is
pending or not.
To suppress lockdep warning, we have to check mddev->del_work while
md_delayed_delete is attached to rdev->del_work, so it is not aligned
to the purpose of flush workquee. So a new workqueue is needed to avoid
the awkward situation, and introduce a new func flush_rdev_wq to flush
the new workqueue after check if there was pending work.
Also like new_dev_store, ADD_NEW_DISK ioctl has the same purpose to flush
workqueue while it holds bdev->bd_mutex, so make the same change applies
to the ioctl to avoid similar lock issue.
And md_delayed_delete actually wants to delete rdev, so rename the function
to rdev_delayed_delete.
Signed-off-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by: Song Liu <songliubraving@fb.com>
Since works (rdev->del_work and mddev->del_work) are queued in md_misc_wq,
then lockdep_map lock is held if either of them are running, then both of
them try to hold kernfs lock by call kobject_del. Then if new_dev_store
or array_state_store are triggered by write to the related sysfs node, so
the write operation gets kernfs lock, but need the lockdep_map because all
of them would trigger flush_workqueue(md_misc_wq) finally, then the same
lockdep_map lock is needed.
To suppress the lockdep warnning, we should flush the workqueue in case the
related work is pending. And several works are attached to md_misc_wq, so
we need to check which work should be checked:
1. for __md_stop_writes, the purpose of call flush workqueue is ensure sync
thread is started if it was starting, so check mddev->del_work is pending
or not since md_start_sync is attached to mddev->del_work.
2. __md_stop flushes md_misc_wq to ensure event_work is done, check the
event_work is enough. Assume raid_{ctr,dtr} -> md_stop -> __md_stop doesn't
need the kernfs lock.
3. both new_dev_store (holds kernfs lock) and ADD_NEW_DISK ioctl (holds the
bdev->bd_mutex) call flush_workqueue to ensure md_delayed_delete has
completed, this case will be handled in next patch.
4. md_open flushes workqueue to ensure the previous md is disappeared, but
it holds bdev->bd_mutex then try to flush workqueue, so it is better to
check mddev->del_work as well to avoid potential lock issue, this will be
done in another patch.
Cc: Coly Li <colyli@suse.de> Reported-by: Coly Li <colyli@suse.de> Signed-off-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by: Song Liu <songliubraving@fb.com>
Jens Axboe [Tue, 12 May 2020 17:35:49 +0000 (11:35 -0600)]
Merge tag 'floppy-for-5.8' of https://github.com/evdenis/linux-floppy into for-5.8/drivers
Floppy patches for 5.8
Cleanups:
- symbolic register names for x86,sparc64,sparc32,powerpc,parisc,m68k
- split of local/global variables for drive,fdc
- UBSAN warning suppress in setup_rw_floppy()
Changes were compile tested on arm, sparc64, powerpc, m68k. Many patches
introduce no binary changes by using defines instead of magic numbers.
The patches were also tested with syzkaller and simple write/read/format
tests on real hardware.
Signed-off-by: Denis Efremov <efremov@linux.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
* tag 'floppy-for-5.8' of https://github.com/evdenis/linux-floppy: (31 commits)
floppy: suppress UBSAN warning in setup_rw_floppy()
floppy: add defines for sizes of cmd & reply buffers of floppy_raw_cmd
floppy: add FD_AUTODETECT_SIZE define for struct floppy_drive_params
floppy: use print_hex_dump() in setup_DMA()
floppy: cleanup: make set_fdc() always set current_drive and current_fd
floppy: cleanup: get rid of current_reqD in favor of current_drive
floppy: make sure to reset all FDCs upon resume()
floppy: cleanup: do not iterate on current_fdc in do_floppy_init()
floppy: cleanup: add a few comments about expectations in certain functions
floppy: cleanup: do not iterate on current_fdc in DMA grab/release functions
floppy: cleanup: make get_fdc_version() not rely on current_fdc anymore
floppy: cleanup: make next_valid_format() not rely on current_drive anymore
floppy: cleanup: make check_wp() not rely on current_{fdc,drive} anymore
floppy: cleanup: make fdc_specify() not rely on current_{fdc,drive} anymore
floppy: cleanup: make fdc_configure() not rely on current_fdc anymore
floppy: cleanup: make perpendicular_mode() not rely on current_fdc anymore
floppy: cleanup: make need_more_output() not rely on current_fdc anymore
floppy: cleanup: make result() not rely on current_fdc anymore
floppy: cleanup: make output_byte() not rely on current_fdc anymore
floppy: cleanup: make wait_til_ready() not rely on current_fdc anymore
...
Denis Efremov [Fri, 1 May 2020 13:44:16 +0000 (16:44 +0300)]
floppy: suppress UBSAN warning in setup_rw_floppy()
UBSAN: array-index-out-of-bounds in drivers/block/floppy.c:1521:45
index 16 is out of range for type 'unsigned char [16]'
Call Trace:
...
setup_rw_floppy+0x5c3/0x7f0
floppy_ready+0x2be/0x13b0
process_one_work+0x2c1/0x5d0
worker_thread+0x56/0x5e0
kthread+0x122/0x170
ret_from_fork+0x35/0x40
This out-of-bounds access is intentional. The command in struct
floppy_raw_cmd may take up the space initially intended for the reply
and the reply count. It is needed for long 82078 commands such as
RESTORE, which takes 17 command bytes. Initial cmd size is not enough
and since struct setup_rw_floppy is a part of uapi we check that
cmd_count is in [0:16+1+16] in raw_cmd_copyin().
The patch adds union with original cmd,reply_count,reply fields and
fullcmd field of equivalent size. The cmd accesses are turned to
fullcmd where appropriate to suppress UBSAN warning.
Denis Efremov [Fri, 1 May 2020 13:44:15 +0000 (16:44 +0300)]
floppy: add defines for sizes of cmd & reply buffers of floppy_raw_cmd
Use FD_RAW_CMD_SIZE, FD_RAW_REPLY_SIZE defines instead of magic numbers
for cmd & reply buffers of struct floppy_raw_cmd. Remove local to
floppy.c MAX_REPLIES define, as it is now FD_RAW_REPLY_SIZE.
FD_RAW_CMD_FULLSIZE added as we allow command to also fill reply_count
and reply fields.
floppy: cleanup: make set_fdc() always set current_drive and current_fd
When called with a negative drive value, set_fdc() would stick to the
current fdc (which was assumed to reflect the current_drive's FDC). We
do not need this anymore as the last call place with a negative value
was just addressed. Let's make this function always set both current_fdc
and current_drive so that there's no more ambiguity. A few comments
stating this were added to a few non-obvious places.
floppy: cleanup: get rid of current_reqD in favor of current_drive
This macro equals -1 and is used as an alternative for current_drive when
calling reschedule_timeout(), which in turn needs to remap it. This only
adds obfuscation, let's simply use current_drive.
In floppy_resume() we don't properly reinitialize all FDCs, instead
we reinitialize the current FDC once per available FDC because value
-1 is passed to user_reset_fdc(). Let's simply save the current drive
and properly reinitialize each FDC.
floppy: cleanup: do not iterate on current_fdc in do_floppy_init()
There's no need to iterate on current_fdc in do_floppy_init() anymore,
in the first case it's only used as an array index to access fdc_state[],
so let's get rid of this confusing assignment. The second case is a bit
trickier because user_reset_fdc() needs to already know current_fdc when
called with drive==-1 due to this call chain:
Note that current_drive is not used in this code part and may even not
match a unit belonging to current_fdc. Instead of passing -1 we can
simply pass the first drive of the FDC being initialized, which is even
cleaner as it will allow the function chain above to consistently assign
both variables.
Willy Tarreau [Tue, 31 Mar 2020 09:40:54 +0000 (11:40 +0200)]
floppy: cleanup: add a few comments about expectations in certain functions
The locking in the driver is far from being obvious, with unlocking
automatically happening at end of operations scheduled by interrupt,
especially for the error paths where one does not necessarily expect
that such an interrupt may be triggered. Let's add a few comments
about what to expect at certain places to avoid misdetecting bugs
which are not.
Willy Tarreau [Tue, 31 Mar 2020 09:40:53 +0000 (11:40 +0200)]
floppy: cleanup: do not iterate on current_fdc in DMA grab/release functions
Both floppy_grab_irq_and_dma() and floppy_release_irq_and_dma() used to
iterate on the global variable while setting up or freeing resources.
Now that they exclusively rely on functions which take the fdc as an
argument, so let's not touch the global one anymore.
Willy Tarreau [Tue, 31 Mar 2020 09:40:47 +0000 (11:40 +0200)]
floppy: cleanup: make perpendicular_mode() not rely on current_fdc anymore
Now the fdc is passed in argument so that the function does not
use current_fdc anymore.
It's worth noting that there's still a single raw_cmd pointer
specific to the current fdc. It may make sense to have one per
fdc in the future. In addition, cont->done() still relies on the
current drive and current raw_cmd.
Willy Tarreau [Tue, 31 Mar 2020 09:40:45 +0000 (11:40 +0200)]
floppy: cleanup: make result() not rely on current_fdc anymore
Now the fdc is passed in argument so that the function does not
use current_fdc anymore.
It's worth noting that there's still a single reply_buffer[] which
will store the result for the current fdc. It may or may not make
sense to implement one buffer per fdc in the future.
Willy Tarreau [Tue, 31 Mar 2020 09:40:37 +0000 (11:40 +0200)]
floppy: use symbolic register names in the sparc32 port
The sparc port used to be forced to rely on numeric register indexes
with their equivalent in comments. Now that they don't depend on the
IO port we can use their symbolic names.
Willy Tarreau [Tue, 31 Mar 2020 09:40:33 +0000 (11:40 +0200)]
floppy: add references to 82077's extra registers
This controller provides extra status registers SRA and SRB as well
as a tape drive register (TDR) and a data rate select register (DSR),
which are referenced in the sparc port, so let's have their symbolic
definitions centralized.
Willy Tarreau [Tue, 31 Mar 2020 09:40:32 +0000 (11:40 +0200)]
floppy: split the base port from the register in I/O accesses
Currently we have architecture-specific fd_inb() and fd_outb() functions
or macros, taking just a port which is in fact made of a base address and
a register. The base address is FDC-specific and derived from the local or
global "fdc" variable through the FD_IOPORT macro used in the base address
calculation.
This change splits this by explicitly passing the FDC's base address and
the register separately to fd_outb() and fd_inb(). It affects the
following archs:
- x86, alpha, mips, powerpc, parisc, arm, m68k:
simple remap of port -> base+reg
- sparc32: use of reg only, since the base address was already masked
out and the FDC controller is known from a static struct.
- sparc64: like x86 for PCI, like sparc32 for 82077
Some archs use inline functions and others macros. This was not
unified in order to minimize the number of changes to review. For the
same reason checkpatch still spews a few warnings about things that
were already there before.
The parisc still uses hard-coded register values and could be cleaned up
by taking the register definitions.
The sparc per-controller inb/outb functions could further be refined
to explicitly take an FDC register instead of a port in argument but it
was not needed yet and may be cleaned later.
Link: https://lore.kernel.org/r/20200331094054.24441-2-w@1wt.eu Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Richard Henderson <rth@twiddle.net> Cc: Matt Turner <mattst88@gmail.com> Cc: Ian Molton <spyro@f2s.com> Cc: Russell King <linux@armlinux.org.uk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Helge Deller <deller@gmx.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: x86@kernel.org Signed-off-by: Willy Tarreau <w@1wt.eu> Signed-off-by: Denis Efremov <efremov@linux.com>
With reference to the NVMeOF Specification (page 44, Figure 38)
discovery log page entry provides address family field. We do set the
transport type field but the adrfam field is not set when using loop
transport and also it doesn't have support in the nvme-cli. So when
reading discovery log page with a loop transport it leads to confusing
output.
As per the spec for adrfam value 254 is reserved for Intra Host
Transport i.e. loopback), we add a required macro in the protocol
header file, set default port disc addr entry's adrfam to
NVMF_ADDR_FAMILY_MAX, and update nvmet_addr_family configfs array for
show/store attribute.
Without this patch, setting adrfam to (ipv4/ipv6/ib/fc/loop/" ") we get
following output for nvme discover command from nvme-cli which is
confusing.
trtype: loop
adrfam: ipv4
trtype: loop
adrfam: ipv6
trtype: loop
adrfam: infiniband
trtype: loop
adrfam: fibre-channel
trtype: loop # ${CFGFS_HOME}/nvmet/ports/1/addr_adrfam = loop
adrfam: pci # <----- pci for loop
trtype: loop # ${CFGFS_HOME}/nvmet/ports/1/addr_adrfam = " "
adrfam: pci # <----- pci for unrecognized
The configfs attributes which are supposed to set when port is disable
such as addr[addrfam|portid|traddr|treq|trsvcid|inline_data_size|trtype]
has repetitive check and generic error message printing.
This patch creates centralize helper to check and print an error
message that also accepts caller as a parameter. This makes error
message easy to parse for the user, removes the duplicate code and
makes it available for futures such scenarios.
Currently nvmet_addr_treq_[store|show]() uses switch and if else
ladder for address transport requirements to string and reverse
mapping. With addtion of the generic nvmet_type_name_map structure
we can get rid of the switch and if else ladder with string
duplication.
Now that we have a generic type to name map for configfs, get rid of
the nvmet_ana_state_names structure and replace it with newly added
nvmet_type_name_map.
Right now nvmet_addr_adrfam_[store|show]() uses switch and if else
ladder for address family to string and reverse mapping which also
repeats the strings in show and store function.
With addition of generic nvmet_type_name_map structure we can now get rid
of the switch and if else ladder and string duplication.
Also, we add a newline in before found label in nvmet_addr_trtype_store()
which keeps goto label code consistent with
nvmet_allowed_hosts_drop_link(), nvmet_port_subsys_drop_link() and
nvmet_ana_group_ana_state_store().
Sagi Grimberg [Fri, 1 May 2020 21:25:45 +0000 (14:25 -0700)]
nvme-tcp: try to send request in queue_rq context
Today, nvme-tcp automatically schedules a send request
to a workqueue context, which is 1 more than we'd need
in case the socket buffer is wide open.
However, because we have async send activity (as a result
of r2t, or write_space callbacks), we need to synchronize
sends from possibly multiple contexts (ideally all running
on the same cpu though).
Thus, we only try to send directly from queue_rq in cases:
1. the send_list is empty
2. we can send it synchronously (i.e. not from the RX path)
3. we run on the same cpu as the queue->io_cpu to avoid
contention on the send operation.
Proposed-by: Mark Wunderlich <mark.wunderlich@intel.com> Signed-off-by: Mark Wunderlich <mark.wunderlich@intel.com> Signed-off-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Sagi Grimberg [Fri, 1 May 2020 21:25:44 +0000 (14:25 -0700)]
nvme-tcp: avoid scheduling io_work if we are already polling
When the user runs polled I/O, we shouldn't have to trigger
the workqueue to generate the receive work upon the .data_ready
upcall. This prevents a redundant context switch when the
application is already polling for completions.
Proposed-by: Mark Wunderlich <mark.wunderlich@intel.com> Signed-off-by: Mark Wunderlich <mark.wunderlich@intel.com> Signed-off-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Weiping Zhang [Sat, 2 May 2020 07:29:41 +0000 (15:29 +0800)]
nvme-pci: align io queue count with allocted nvme_queue in nvme_probe
Since commit 147b27e4bd08 ("nvme-pci: allocate device queues storage
space at probe"), nvme_alloc_queue does not alloc the nvme queues
itself anymore.
If the write/poll_queues module parameters are changed at runtime to
values larger than the number of allocated queues in nvme_probe,
nvme_alloc_queue will access unallocated memory.
Add a new nr_allocated_queues member to struct nvme_dev to record how
many queues were alloctated in nvme_probe to avoid using more than the
allocated queues after a reset following a change to the
write/poll_queues module parameters.
Also add nr_write_queues and nr_poll_queues members to allow refreshing
the number of write and poll queues based on a change to the module
parameters when resetting the controller.
Fixes: 147b27e4bd08 ("nvme-pci: allocate device queues storage space at probe") Signed-off-by: Weiping Zhang <zhangweiping@didiglobal.com> Reviewed-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
[hch: add nvme_max_io_queues, update the commit message] Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Tue, 28 Apr 2020 14:21:56 +0000 (07:21 -0700)]
nvme-pci: remove volatile cqes
The completion queue entry is not volatile once the phase is confirmed.
Remove the volatile keywords and check the phase using the appropriate
READ_ONCE() accessor, allowing the compiler to optimize the remaining
completion path.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Wed, 29 Apr 2020 20:31:23 +0000 (05:31 +0900)]
nvme: flush scan work on passthrough commands
If a passthrough command causes the namespace inventory or capabilities
to change, flush the scan work that handles these changes so the driver
synchronizes with the user command's effects before returning the result
to user space.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
When CONFIG_ARCH_NO_SG_CHAIN is set, op->sgl[0] cannot be dereferenced,
as gcc-10 now points out:
drivers/nvme/host/fc.c: In function 'nvme_fc_init_request':
drivers/nvme/host/fc.c:1774:29: warning: array subscript 0 is outside the bounds of an interior zero-length array 'struct scatterlist[0]' [-Wzero-length-bounds]
1774 | op->op.fcp_req.first_sgl = &op->sgl[0];
| ^~~~~~~~~~~
drivers/nvme/host/fc.c:98:21: note: while referencing 'sgl'
98 | struct scatterlist sgl[NVME_INLINE_SG_CNT];
| ^~~
I don't know if this is a legitimate warning or a false-positive.
If this is just a false alarm, the warning is easily suppressed
by interpreting the array as a pointer.
Fixes: b1ae1a238900 ("nvme-fc: Avoid preallocating big SGL for data") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Add support for detecting capacity changes on nvmet blockdev and file
backed namespaces. This allows for emulating and testing online resizing
of nvme devices and filesystems on top.
Signed-off-by: Anthony Iliopoulos <ailiop@suse.com>
[chaitanya: Fix comments posted on V1] Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
[hch: reuse code a bit more] Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Thu, 9 Apr 2020 16:09:08 +0000 (09:09 -0700)]
nvme: consolodate io settings
The stream parameters indicating optimal io settings were just getting
overwritten later. Rearrange the settings so the streams parameters can
be preserved if provided.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Thu, 9 Apr 2020 16:09:07 +0000 (09:09 -0700)]
nvme: revalidate namespace stream parameters
The stream parameters are based on the currently formatted logical block
size. Recheck these parameters on namespace revalidation so the
registered constraints will be accurate.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Thu, 9 Apr 2020 16:09:06 +0000 (09:09 -0700)]
nvme: consolidate chunk_sectors settings
Move the quirked chunk_sectors setting to the same location as noiob so
one place registers this setting. And since the noiob value is only used
locally, remove the member from struct nvme_ns.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Thu, 9 Apr 2020 16:09:04 +0000 (09:09 -0700)]
nvme-multipath: set bdi capabilities once
The queues' backing device info capabilities don't change with each
namespace revalidation. Set it only when each path's request_queue
is initially added to a multipath queue.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Thu, 9 Apr 2020 16:09:01 +0000 (09:09 -0700)]
nvme: always search for namespace head
Even if a namespace reports it is not capable of sharing, search the
subsystem for a matching namespace head. If found, the driver should
reject that namespace since it's coming from an invalid configuration.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Thu, 9 Apr 2020 16:09:00 +0000 (09:09 -0700)]
nvme: release namespace head reference on error
If a namespace identification does not match the subsystem's head for
that NSID, release the reference that was taken when the matching head
was initially found.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keith Busch [Thu, 9 Apr 2020 16:08:59 +0000 (09:08 -0700)]
nvme: unlink head after removing last namespace
The driver had been unlinking the namespace head from the subsystem's
list only after the last reference was released, and outside of the
list's subsys->lock protection.
There is no reason to track an empty head, so unlink the entry from the
subsystem's list when the last namespace using that head is removed and
with the mutex lock protecting the list update. The next namespace to
attach reusing the previous NSID will allocate a new head rather than
find the old head with mismatched identifiers.
Signed-off-by: Keith Busch <kbusch@kernel.org> Reviewed-by: Sagi Grimberg <sagi@grimberg.me> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>