Skip to contentSkip to conversation

Followup single fix for block IO core pull

GIT PULL Jens Axboe 2 participants 4 messages Apr 17, 2015

lore
Share & tools
lore

Jens Axboe

author
#0 lore raw

Hi Linus,

A commit in the previous pull request introduce a regression. So far only observed on qemu-sparc64, but it's a general bug. Please pull this single fix to rectify that, thanks.

git://git.kernel.dk/linux-block.git for-linus

----------------------------------------------------------------
Jens Axboe (1):
blk-mq: fix iteration of busy bitmap

 block/blk-mq.c         | 6 +++---
 include/linux/blk-mq.h | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)
Signature · 1 line
Jens Axboe
#1 lore raw
Quoted 5 lines · Jens Axboe

Hi Linus,

A commit in the previous pull request introduce a regression. So far only observed on qemu-sparc64, but it's a general bug. Please pull this single fix to rectify that, thanks.

I hit the same bug on two x86 boxes, bare-metal. Reverting the commit Guenter pointed out made them boot again, so definitely not sparc/qemu specific in any way.

I suspect the only reason more people didn't see it is that not everyone is running with the mq-by-default config option yet ?

Dave

author
#2 lore raw
Quoted 7 lines · Dave Jones
Quoted 5 lines · Jens Axboe

Hi Linus,

A commit in the previous pull request introduce a regression. So far only observed on qemu-sparc64, but it's a general bug. Please pull this single fix to rectify that, thanks.

Quoted 6 lines

I hit the same bug on two x86 boxes, bare-metal. Reverting the commit Guenter pointed out made them boot again, so definitely not sparc/qemu specific in any way.

I suspect the only reason more people didn't see it is that not everyone is running with the mq-by-default config option yet ?

It is puzzling. I ran it on several boxes, both multi queue and not, and both single mq queues and multiple. But the code is clearly wrong. So yeah, it's not sparc64 specific in any way. My initial thought was that this was another issue related to sparse CPU ids, but that's not the case. It's a plain bug, unfortunately.

Signature · 1 line
Jens Axboe
author
#3 lore raw
Quoted 13 lines · Jens Axboe
Quoted 12 lines · Dave Jones
Quoted 4 lines · Jens Axboe

Hi Linus,

A commit in the previous pull request introduce a regression. So far only observed on qemu-sparc64, but it's a general bug. Please pull

this

Quoted 1 line

single fix to rectify that, thanks.

I hit the same bug on two x86 boxes, bare-metal. Reverting the commit Guenter pointed out made them boot again, so definitely not sparc/qemu specific in any way.

Quoted 8 lines
Quoted 2 lines

I suspect the only reason more people didn't see it is that not everyone is running with the mq-by-default config option yet ?

It is puzzling. I ran it on several boxes, both multi queue and not, and both single mq queues and multiple. But the code is clearly wrong. So yeah, it's not sparc64 specific in any way. My initial thought was that this was another issue related to sparse CPU ids, but that's not the case. It's a plain bug, unfortunately.

It's the pre morning coffee issue, looks like. All my test boxes end up having >= map->bits_per_word CPUs, so the test cases never ran into the round down issue.

Signature · 1 line
Jens Axboe