On Thu, Nov 3, 2016 at 5:42 AM, Kamal Mostafa <ka...@canonical.com> wrote: > Ming Lei comment #2 says you're the author of this patch to the hio > driver: > > +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0)) > + blk_queue_split(q, &bio, q->bio_split); > +#endif > + > > Can you provide us with a short explanation for the git log, and also > your Signed-off-by line for that patch?
Sure, please see the attachment. > > -- > You received this bug notification because you are subscribed to the bug > report. > https://bugs.launchpad.net/bugs/1638700 > > Title: > hio: SSD data corruption under stress test > > Status in linux package in Ubuntu: > In Progress > Status in linux source package in Xenial: > In Progress > Status in linux source package in Yakkety: > In Progress > Status in linux source package in Zesty: > In Progress > > Bug description: > {forward from James Troup}: > > Just to followup to this with a little more information, we have now > reproduced this in the following scenarios: > > * Ubuntu kernel 4.4 (i.e. 16.04) and kernel 4.8 (i.e. HWE-Y) > * With and without Bcache involved > * With both XFS and ext4 > * With HIO driver versions 2.1.0-23 and 2.1.0-25 > * With HIO Firmware 640 and 650 > * With and without the following two patches > - > https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=7290fa97b945c288d8dd8eb8f284b98cb495b35b > - > https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=901a3142db778ddb9ed6a9000ce8e5b0f66c48ba > > In all cases, we applied the following two patches in order to get hio > to build at all with a 4.4 or later kernel: > > > https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=0abbb90372847caeeedeaa9db0f21e05ad8e9c74 > > https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=a0705c5ff3d12fc31f18f5d3c8589eaaed1aa577 > > We've confirmed that we can reproduce the corruption on any machine in > Tele2's Vienna facility. > > We've confirmed that, other than 1 machine, the 'hio_info' command > says the health is 'OK'. > > Our most common reproducer is one of two scenarios: > > a) http://paste.ubuntu.com/23405150/ > > b) http://paste.ubuntu.com/23405234/ > > In the last example, it's possible to see corruption faster by > increasing the 'count' argument to dd and avoid it by lowering it. > e.g. on the machine I'm currently testing on count=52450 doesn't > appear to show corruption, but a count of even 53000 would show it > immediately every time. > > I hope this helps - please let us know what further information we can > provide to debug this problem. > > To manage notifications about this bug go to: > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1638700/+subscriptions ** Patch added: "0001-hio.c-splitting-bio-in-the-entry-of-.make_request_fn.patch" https://bugs.launchpad.net/bugs/1638700/+attachment/4771506/+files/0001-hio.c-splitting-bio-in-the-entry-of-.make_request_fn.patch -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu. https://bugs.launchpad.net/bugs/1638700 Title: hio: SSD data corruption under stress test Status in linux package in Ubuntu: In Progress Status in linux source package in Xenial: In Progress Status in linux source package in Yakkety: In Progress Status in linux source package in Zesty: In Progress Bug description: {forward from James Troup}: Just to followup to this with a little more information, we have now reproduced this in the following scenarios: * Ubuntu kernel 4.4 (i.e. 16.04) and kernel 4.8 (i.e. HWE-Y) * With and without Bcache involved * With both XFS and ext4 * With HIO driver versions 2.1.0-23 and 2.1.0-25 * With HIO Firmware 640 and 650 * With and without the following two patches - https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=7290fa97b945c288d8dd8eb8f284b98cb495b35b - https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=901a3142db778ddb9ed6a9000ce8e5b0f66c48ba In all cases, we applied the following two patches in order to get hio to build at all with a 4.4 or later kernel: https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=0abbb90372847caeeedeaa9db0f21e05ad8e9c74 https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/commit/?id=a0705c5ff3d12fc31f18f5d3c8589eaaed1aa577 We've confirmed that we can reproduce the corruption on any machine in Tele2's Vienna facility. We've confirmed that, other than 1 machine, the 'hio_info' command says the health is 'OK'. Our most common reproducer is one of two scenarios: a) http://paste.ubuntu.com/23405150/ b) http://paste.ubuntu.com/23405234/ In the last example, it's possible to see corruption faster by increasing the 'count' argument to dd and avoid it by lowering it. e.g. on the machine I'm currently testing on count=52450 doesn't appear to show corruption, but a count of even 53000 would show it immediately every time. I hope this helps - please let us know what further information we can provide to debug this problem. To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1638700/+subscriptions -- Mailing list: https://launchpad.net/~kernel-packages Post to : kernel-packages@lists.launchpad.net Unsubscribe : https://launchpad.net/~kernel-packages More help : https://help.launchpad.net/ListHelp
