block/floppy: fix contended case in floppy_queue_rq()
authorJiri Kosina <jkosina@suse.cz>
Tue, 26 May 2020 09:49:18 +0000 (11:49 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 17 Jun 2020 14:40:38 +0000 (16:40 +0200)
commit29ed45653bec27b1a677b5af551ef5525e3333d1
tree4db66be0e1f2daf95b12bb100f1b119c02363e09
parent9ec34f441ed480a121775dc77fa972f9a311cb5c
block/floppy: fix contended case in floppy_queue_rq()

commit 263c61581a38d0a5ad1f5f4a9143b27d68caeffd upstream.

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.

Fixes: a9f38e1dec107a ("floppy: convert to blk-mq")
Cc: stable@vger.kernel.org
Tested-by: Libor Pechacek <lpechacek@suse.cz>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/block/floppy.c