Re: [PULL v2] hw/nvme fixes

On 8/8/23 06:35, Klaus Jensen wrote: From: Klaus Jensen Hi, There was a small typo in the last pull. This replaces it. The following changes since commit 0450cf08976f9036feaded438031b4cba94f6452: Merge tag 'fixes-pull-request' ofhttps://gitlab.com/marcandre.lureau/qemu into staging (2023

Re: [PATCH 2/2] block: change reqs_lock to QemuMutex

On Tue, Aug 08, 2023 at 11:58:52AM -0400, Stefan Hajnoczi wrote: > CoMutex has poor performance when lock contention is high. The tracked > requests list is accessed frequently and performance suffers in QEMU > multi-queue block layer scenarios. > > It is not necessary to use CoMutex for the reque

Re: [PATCH 1/2] block: minimize bs->reqs_lock section in tracked_request_end()

On Tue, Aug 08, 2023 at 11:58:51AM -0400, Stefan Hajnoczi wrote: > Signed-off-by: Stefan Hajnoczi > --- > block/io.c | 8 +++- > 1 file changed, 7 insertions(+), 1 deletion(-) Reviewed-by: Eric Blake > > diff --git a/block/io.c b/block/io.c > index 055fcf7438..85d5176256 100644 > --- a/bl

Re: [PATCH] block-migration: Ensure we don't crash during migration cleanup

On Mon, Jul 31, 2023 at 05:33:38PM -0300, Fabiano Rosas wrote: > We can fail the blk_insert_bs() at init_blk_migration(), leaving the > BlkMigDevState without a dirty_bitmap and BlockDriverState. Account > for the possibly missing elements when doing cleanup. > > Fix the following crashes: > > Th

[PATCH 2/2] block: change reqs_lock to QemuMutex

CoMutex has poor performance when lock contention is high. The tracked requests list is accessed frequently and performance suffers in QEMU multi-queue block layer scenarios. It is not necessary to use CoMutex for the requests lock. The lock is always released across coroutine yield operations. It

[PATCH 0/2] block: change reqs_lock to QemuMutex

As part of the ongoing multi-queue QEMU block layer work, I found that CoMutex reqs_lock scales poorly when more IOThreads are added. These patches double IOPS in the 4 IOThreads randread benchmark that I have been running with my out-of-tree virtio-blk-iothread-vq-mapping branch (https://gitlab.co

[PATCH 1/2] block: minimize bs->reqs_lock section in tracked_request_end()

Signed-off-by: Stefan Hajnoczi --- block/io.c | 8 +++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/block/io.c b/block/io.c index 055fcf7438..85d5176256 100644 --- a/block/io.c +++ b/block/io.c @@ -593,8 +593,14 @@ static void coroutine_fn tracked_request_end(BdrvTrackedRequ

[PATCH 2/2] hw/nvme: fix null pointer access in ruh update

From: Klaus Jensen The Reclaim Unit Update operation in I/O Management Receive does not verify the presence of a configured endurance group prior to accessing it. Fix this. Cc: qemu-sta...@nongnu.org Fixes: 73064edfb864 ("hw/nvme: flexible data placement emulation") Signed-off-by: Klaus Jensen

[PATCH 0/2] hw/nvme: two fixes

From: Klaus Jensen Fix two potential accesses to null pointers. Klaus Jensen (2): hw/nvme: fix null pointer access in directive receive hw/nvme: fix null pointer access in ruh update hw/nvme/ctrl.c | 10 -- 1 file changed, 8 insertions(+), 2 deletions(-) -- 2.41.0

[PATCH 1/2] hw/nvme: fix null pointer access in directive receive

From: Klaus Jensen nvme_directive_receive() does not check if an endurance group has been configured (set) prior to testing if flexible data placement is enabled or not. Fix this. Cc: qemu-sta...@nongnu.org Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1815 Fixes: 73064edfb864 ("hw/nv

[PULL v2] hw/nvme fixes

From: Klaus Jensen Hi, There was a small typo in the last pull. This replaces it. The following changes since commit 0450cf08976f9036feaded438031b4cba94f6452: Merge tag 'fixes-pull-request' of https://gitlab.com/marcandre.lureau/qemu into staging (2023-08-07 13:55:00 -0700) are available i

Re: [PATCH] block-migration: Ensure we don't crash during migration cleanup

added Kevin and Hanna for block, since this seems still untouched? Thanks, Claudio On 7/31/23 22:33, Fabiano Rosas wrote: > We can fail the blk_insert_bs() at init_blk_migration(), leaving the > BlkMigDevState without a dirty_bitmap and BlockDriverState. Account > for the possibly missing elemen