| Message ID | 20220801215043.243173-1-alexander.mikhalitsyn@virtuozzo.com |
|---|---|
| State | New |
| Headers | show |
| Series | [SRU,J,master-next] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency | expand |
On Tue, Aug 02, 2022 at 12:50:43AM +0300, Alexander Mikhalitsyn wrote: > BugLink: https://bugs.launchpad.net/bugs/1967924 > BugLink: https://bugs.launchpad.net/bugs/1857257 > > BugLink: https://bugs.launchpad.net/bugs/1967924 > > Right now we have a fix: > b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > in master branch of Jammy kernel, but only formaly. Because these kernels compiled without > CONFIG_AUFS_FS set, so this fix just disabled. There is no need to make it dependent on > CONFIG_AUFS_FS option, because in all cases we have mm/prfile.c compiled-in. > > Cc: Andrei Vagin <avagin@gmail.com> > Cc: Adrian Reber <areber@redhat.com> > Cc: Stefan Bader <stefan.bader@canonical.com> > Cc: Kleber Sacilotto de Souza <kleber.souza@canonical.com> > Cc: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> > Cc: Andrea Righi <andrea.righi@canonical.com> > > Fixes: b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > Signed-off-by: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com> It's true that we don't need this in Jammy, but we need it in focal for the hwe kernels, where we still need to support AUFS. So, we decided to keep it also in Jammy, even if, as you correctly mentioned, it's just formal change, since this patch is a no-op with AUFS disabled, but it helps to maintain the code especially when we need to release new hwe kernels. -Andrea
On Tue, 2 Aug 2022 09:05:31 +0200 Andrea Righi <andrea.righi@canonical.com> wrote: > On Tue, Aug 02, 2022 at 12:50:43AM +0300, Alexander Mikhalitsyn wrote: > > BugLink: https://bugs.launchpad.net/bugs/1967924 > > BugLink: https://bugs.launchpad.net/bugs/1857257 > > > > BugLink: https://bugs.launchpad.net/bugs/1967924 > > > > Right now we have a fix: > > b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > > in master branch of Jammy kernel, but only formaly. Because these kernels compiled without > > CONFIG_AUFS_FS set, so this fix just disabled. There is no need to make it dependent on > > CONFIG_AUFS_FS option, because in all cases we have mm/prfile.c compiled-in. > > > > Cc: Andrei Vagin <avagin@gmail.com> > > Cc: Adrian Reber <areber@redhat.com> > > Cc: Stefan Bader <stefan.bader@canonical.com> > > Cc: Kleber Sacilotto de Souza <kleber.souza@canonical.com> > > Cc: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> > > Cc: Andrea Righi <andrea.righi@canonical.com> > > > > Fixes: b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > > Signed-off-by: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com> Hi Andrea, > > It's true that we don't need this in Jammy, but we need it in focal for > the hwe kernels, where we still need to support AUFS. So, we decided to I'm sorry, but I don't understand your point. This patch *completely* independent of CONFIG_AUFS_FS it only relies on the *one* small piece of code from aufs: https://github.com/sfjro/aufs5-linux/commit/f8a27912904bdc40a27f434a9b6d19f56e7fb3b6 which is present in the master-next branch. And this code is always compiled-in regardless of the CONFIG_AUFS_FS kernel. But for this branch we need to remove this check because it makes this fix totally useless and users are still facing issues with overlayfs. > keep it also in Jammy, even if, as you correctly mentioned, it's just > formal change, since this patch is a no-op with AUFS disabled, but it > helps to maintain the code especially when we need to release new hwe > kernels. I've prepared corresponding separate (!) patches for hwe-5.17 branch. Please take a look. > > -Andrea Regards, Alex
On Tue, Aug 02, 2022 at 11:04:23AM +0300, Alexander Mikhalitsyn wrote: > On Tue, 2 Aug 2022 09:05:31 +0200 > Andrea Righi <andrea.righi@canonical.com> wrote: > > > On Tue, Aug 02, 2022 at 12:50:43AM +0300, Alexander Mikhalitsyn wrote: > > > BugLink: https://bugs.launchpad.net/bugs/1967924 > > > BugLink: https://bugs.launchpad.net/bugs/1857257 > > > > > > BugLink: https://bugs.launchpad.net/bugs/1967924 > > > > > > Right now we have a fix: > > > b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > > > in master branch of Jammy kernel, but only formaly. Because these kernels compiled without > > > CONFIG_AUFS_FS set, so this fix just disabled. There is no need to make it dependent on > > > CONFIG_AUFS_FS option, because in all cases we have mm/prfile.c compiled-in. > > > > > > Cc: Andrei Vagin <avagin@gmail.com> > > > Cc: Adrian Reber <areber@redhat.com> > > > Cc: Stefan Bader <stefan.bader@canonical.com> > > > Cc: Kleber Sacilotto de Souza <kleber.souza@canonical.com> > > > Cc: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> > > > Cc: Andrea Righi <andrea.righi@canonical.com> > > > > > > Fixes: b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > > > Signed-off-by: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com> > > Hi Andrea, > > > > > It's true that we don't need this in Jammy, but we need it in focal for > > the hwe kernels, where we still need to support AUFS. So, we decided to > > I'm sorry, but I don't understand your point. > > This patch *completely* independent of CONFIG_AUFS_FS it only relies on > the *one* small piece of code from aufs: > https://github.com/sfjro/aufs5-linux/commit/f8a27912904bdc40a27f434a9b6d19f56e7fb3b6 > which is present in the master-next branch. > > And this code is always compiled-in regardless of the CONFIG_AUFS_FS kernel. > > But for this branch we need to remove this check because it makes this fix totally useless and users are still facing issues with overlayfs. Does it mean that this fix isn't really fixing the issue with overlayfs? In this case we should definitely drop it, but in this case I would rephrase the commit description, mentioning that the fix is not correct. > > > keep it also in Jammy, even if, as you correctly mentioned, it's just > > formal change, since this patch is a no-op with AUFS disabled, but it > > helps to maintain the code especially when we need to release new hwe > > kernels. > > I've prepared corresponding separate (!) patches for hwe-5.17 branch. Please take a look. Oh, I see now that you sent a patch set against hwe-5.17. We may also want to fix this properly in hwe-5.15 as well. Do you have a patch for 5.15 already? Thanks for the clarification, -Andrea
On Tue, 2 Aug 2022 10:12:55 +0200 Andrea Righi <andrea.righi@canonical.com> wrote: > On Tue, Aug 02, 2022 at 11:04:23AM +0300, Alexander Mikhalitsyn wrote: > > On Tue, 2 Aug 2022 09:05:31 +0200 > > Andrea Righi <andrea.righi@canonical.com> wrote: > > > > > On Tue, Aug 02, 2022 at 12:50:43AM +0300, Alexander Mikhalitsyn wrote: > > > > BugLink: https://bugs.launchpad.net/bugs/1967924 > > > > BugLink: https://bugs.launchpad.net/bugs/1857257 > > > > > > > > BugLink: https://bugs.launchpad.net/bugs/1967924 > > > > > > > > Right now we have a fix: > > > > b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > > > > in master branch of Jammy kernel, but only formaly. Because these kernels compiled without > > > > CONFIG_AUFS_FS set, so this fix just disabled. There is no need to make it dependent on > > > > CONFIG_AUFS_FS option, because in all cases we have mm/prfile.c compiled-in. > > > > > > > > Cc: Andrei Vagin <avagin@gmail.com> > > > > Cc: Adrian Reber <areber@redhat.com> > > > > Cc: Stefan Bader <stefan.bader@canonical.com> > > > > Cc: Kleber Sacilotto de Souza <kleber.souza@canonical.com> > > > > Cc: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> > > > > Cc: Andrea Righi <andrea.righi@canonical.com> > > > > > > > > Fixes: b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") > > > > Signed-off-by: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com> > > > > Hi Andrea, > > > > > > > > It's true that we don't need this in Jammy, but we need it in focal for > > > the hwe kernels, where we still need to support AUFS. So, we decided to > > > > I'm sorry, but I don't understand your point. > > > > This patch *completely* independent of CONFIG_AUFS_FS it only relies on > > the *one* small piece of code from aufs: > > https://github.com/sfjro/aufs5-linux/commit/f8a27912904bdc40a27f434a9b6d19f56e7fb3b6 > > which is present in the master-next branch. > > > > And this code is always compiled-in regardless of the CONFIG_AUFS_FS kernel. > > > > But for this branch we need to remove this check because it makes this fix totally useless and users are still facing issues with overlayfs. > > Does it mean that this fix isn't really fixing the issue with overlayfs? Yes, because Ubuntu Jammy kernels was built without CONFIG_AUFS_FS, so this patch is just no-op. Example: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1967924/comments/21 https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1967924/comments/24 > In this case we should definitely drop it, but in this case I would > rephrase the commit description, mentioning that the fix is not correct. I believe that we just need to apply this fixup patch to the original one and squash it to one commit. > > > > > > keep it also in Jammy, even if, as you correctly mentioned, it's just > > > formal change, since this patch is a no-op with AUFS disabled, but it > > > helps to maintain the code especially when we need to release new hwe > > > kernels. > > > > I've prepared corresponding separate (!) patches for hwe-5.17 branch. Please take a look. > > Oh, I see now that you sent a patch set against hwe-5.17. We may also > want to fix this properly in hwe-5.15 as well. Do you have a patch for > 5.15 already? I can't see hwe-5.15 branch for jammy here: https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy/ Or you mean hwe-5.15 focal? Of course, I'm ready to send patches for focal too. I've just started working on Jammy. After getting review I'm ready to send follow up patches for focal branches too. > > Thanks for the clarification, > -Andrea Thanks for your attention to this bug! It's a real pain for CRIU users ;-) offtop: for some reason my emails getting posted on https://lists.ubuntu.com/archives/kernel-team/2022-August/thread.html very slowly. Should I ask lists admins about that? Or possibly, the problem is that each my email should be checked manually by list admin because of spam? I wonder If my email can be added to trusted senders, I've no plan to send marketing to this lists right now ;) Regards, Alex
On Tue, Aug 02, 2022 at 11:42:55AM +0300, Alexander Mikhalitsyn wrote: > > > > It's true that we don't need this in Jammy, but we need it in focal for > > > > the hwe kernels, where we still need to support AUFS. So, we decided to > > > > > > I'm sorry, but I don't understand your point. > > > > > > This patch *completely* independent of CONFIG_AUFS_FS it only relies on > > > the *one* small piece of code from aufs: > > > https://github.com/sfjro/aufs5-linux/commit/f8a27912904bdc40a27f434a9b6d19f56e7fb3b6 > > > which is present in the master-next branch. > > > > > > And this code is always compiled-in regardless of the CONFIG_AUFS_FS kernel. > > > > > > But for this branch we need to remove this check because it makes this fix totally useless and users are still facing issues with overlayfs. > > > > Does it mean that this fix isn't really fixing the issue with overlayfs? > > Yes, because Ubuntu Jammy kernels was built without CONFIG_AUFS_FS, so this patch is just no-op. > Example: > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1967924/comments/21 > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1967924/comments/24 > > > In this case we should definitely drop it, but in this case I would > > rephrase the commit description, mentioning that the fix is not correct. > > I believe that we just need to apply this fixup patch to the original one and squash it to one commit. So, IIUC we need the chunk of code from AUFS to introduce vm_prfile, then this fix needs to be changed to use vm_prfile (like we do in focal) even without AUFS. Basically this is what you've done with hwe-5.17, right? > > > > > > > > > > keep it also in Jammy, even if, as you correctly mentioned, it's just > > > > formal change, since this patch is a no-op with AUFS disabled, but it > > > > helps to maintain the code especially when we need to release new hwe > > > > kernels. > > > > > > I've prepared corresponding separate (!) patches for hwe-5.17 branch. Please take a look. > > > > Oh, I see now that you sent a patch set against hwe-5.17. We may also > > want to fix this properly in hwe-5.15 as well. Do you have a patch for > > 5.15 already? > > I can't see hwe-5.15 branch for jammy here: > https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy/ > > Or you mean hwe-5.15 focal? Of course, I'm ready to send patches for focal too. I've just started working on Jammy. After getting review I'm ready to send follow up patches for focal branches too. Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no rush to have the focal patch, I was just pointing out that we need to fix also this kernel to make everything aligned/consistent. > > > > > Thanks for the clarification, > > -Andrea > > Thanks for your attention to this bug! It's a real pain for CRIU users ;-) > > offtop: for some reason my emails getting posted on > https://lists.ubuntu.com/archives/kernel-team/2022-August/thread.html > very slowly. Should I ask lists admins about that? Or possibly, the problem is that > each my email should be checked manually by list admin because of spam? > I wonder If my email can be added to trusted senders, I've no plan to send marketing to this lists right now ;) Hm... not sure why you're experiencing this problem, I'll ask around. :) Thanks, -Andrea
On 02.08.22 14:22, Andrea Righi wrote: > On Tue, Aug 02, 2022 at 11:42:55AM +0300, Alexander Mikhalitsyn wrote: >>>>> It's true that we don't need this in Jammy, but we need it in focal for >>>>> the hwe kernels, where we still need to support AUFS. So, we decided to >>>> >>>> I'm sorry, but I don't understand your point. >>>> >>>> This patch *completely* independent of CONFIG_AUFS_FS it only relies on >>>> the *one* small piece of code from aufs: >>>> https://github.com/sfjro/aufs5-linux/commit/f8a27912904bdc40a27f434a9b6d19f56e7fb3b6 >>>> which is present in the master-next branch. >>>> >>>> And this code is always compiled-in regardless of the CONFIG_AUFS_FS kernel. >>>> >>>> But for this branch we need to remove this check because it makes this fix totally useless and users are still facing issues with overlayfs. >>> >>> Does it mean that this fix isn't really fixing the issue with overlayfs? >> >> Yes, because Ubuntu Jammy kernels was built without CONFIG_AUFS_FS, so this patch is just no-op. >> Example: >> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1967924/comments/21 >> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1967924/comments/24 >> >>> In this case we should definitely drop it, but in this case I would >>> rephrase the commit description, mentioning that the fix is not correct. >> >> I believe that we just need to apply this fixup patch to the original one and squash it to one commit. > > So, IIUC we need the chunk of code from AUFS to introduce vm_prfile, > then this fix needs to be changed to use vm_prfile (like we do in focal) > even without AUFS. > > Basically this is what you've done with hwe-5.17, right? > >> >>> >>>> >>>>> keep it also in Jammy, even if, as you correctly mentioned, it's just >>>>> formal change, since this patch is a no-op with AUFS disabled, but it >>>>> helps to maintain the code especially when we need to release new hwe >>>>> kernels. >>>> >>>> I've prepared corresponding separate (!) patches for hwe-5.17 branch. Please take a look. >>> >>> Oh, I see now that you sent a patch set against hwe-5.17. We may also >>> want to fix this properly in hwe-5.15 as well. Do you have a patch for >>> 5.15 already? >> >> I can't see hwe-5.15 branch for jammy here: >> https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy/ >> >> Or you mean hwe-5.15 focal? Of course, I'm ready to send patches for focal too. I've just started working on Jammy. After getting review I'm ready to send follow up patches for focal branches too. > > Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > rush to have the focal patch, I was just pointing out that we need to > fix also this kernel to make everything aligned/consistent. > >> >>> >>> Thanks for the clarification, >>> -Andrea >> >> Thanks for your attention to this bug! It's a real pain for CRIU users ;-) >> >> offtop: for some reason my emails getting posted on >> https://lists.ubuntu.com/archives/kernel-team/2022-August/thread.html >> very slowly. Should I ask lists admins about that? Or possibly, the problem is that >> each my email should be checked manually by list admin because of spam? >> I wonder If my email can be added to trusted senders, I've no plan to send marketing to this lists right now ;) > > Hm... not sure why you're experiencing this problem, I'll ask around. :) You get moderated because you are not subscribed to the mailing list. You can be added as an exception but you could as well subscribe for now and unsubscribe later. Would have the advantage that you see replies even if people forget to reply to all. ;) -Stefan > > Thanks, > -Andrea
On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: ... > > > > Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > > rush to have the focal patch, I was just pointing out that we need to > > fix also this kernel to make everything aligned/consistent. > > Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) Alright, after all the clarifications I change my NACK to an ACK: Acked-by: Andrea Righi <andrea.righi@canonical.com> The other kernels that require this fix are: kinetic/linux: - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/kinetic jammy/linux: - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy If the fix is applied to both these kernels all the derivatives will inherit the fix automatically (there's also hwe-5.17, but you already provided a fix for that). Thanks, -Andrea
On 03.08.22 08:04, Andrea Righi wrote: > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > ... >>> >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no >>> rush to have the focal patch, I was just pointing out that we need to >>> fix also this kernel to make everything aligned/consistent. >> >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > Alright, after all the clarifications I change my NACK to an ACK: Sorry with the amount of things incoming this does not fly well. It it better to resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here are the target kernel names and if that is omitted it is the primary "linux" source. jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is the parent of all 5.17 based kernels). Any other HWE kernels normally are derived from a primary kernel. So they inherit the changes automatically. -Stefan > > Acked-by: Andrea Righi <andrea.righi@canonical.com> > > The other kernels that require this fix are: > > kinetic/linux: > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/kinetic > > jammy/linux: > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy > > If the fix is applied to both these kernels all the derivatives will > inherit the fix automatically (there's also hwe-5.17, but you already > provided a fix for that). > > Thanks, > -Andrea
On Wed, 3 Aug 2022 09:24:17 +0200 Stefan Bader <stefan.bader@canonical.com> wrote: > On 03.08.22 08:04, Andrea Righi wrote: > > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > > ... > >>> > >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > >>> rush to have the focal patch, I was just pointing out that we need to > >>> fix also this kernel to make everything aligned/consistent. > >> > >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > > > Alright, after all the clarifications I change my NACK to an ACK: > > Sorry with the amount of things incoming this does not fly well. It it better to > resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here > are the target kernel names and if that is omitted it is the primary "linux" source. > jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is > the parent of all 5.17 based kernels). Any other HWE kernels normally are > derived from a primary kernel. So they inherit the changes automatically. Dear friends, if I understand you correctly, I need to resend only [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency and recheck that it fits well for Jammy and Kinetic non-hwe kernels. I'll do that and resend this patch with detailed info about applicable branches. For hwe kernels I'll check too. Regards, Alex > > -Stefan > > > > > Acked-by: Andrea Righi <andrea.righi@canonical.com> > > > > The other kernels that require this fix are: > > > > kinetic/linux: > > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/kinetic > > > > jammy/linux: > > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy > > > > If the fix is applied to both these kernels all the derivatives will > > inherit the fix automatically (there's also hwe-5.17, but you already > > provided a fix for that). > > > > Thanks, > > -Andrea >
On Wed, 3 Aug 2022 09:24:17 +0200 Stefan Bader <stefan.bader@canonical.com> wrote: > On 03.08.22 08:04, Andrea Righi wrote: > > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > > ... > >>> > >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > >>> rush to have the focal patch, I was just pointing out that we need to > >>> fix also this kernel to make everything aligned/consistent. > >> > >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > > > Alright, after all the clarifications I change my NACK to an ACK: > > Sorry with the amount of things incoming this does not fly well. It it better to > resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here > are the target kernel names and if that is omitted it is the primary "linux" source. > jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is > the parent of all 5.17 based kernels). Any other HWE kernels normally are > derived from a primary kernel. So they inherit the changes automatically. Dear friends, I've rechecked different kernels and this is full summary: ============================ * jammy/linux - hwe-5.17 * kinetic/linux - master-next these branches has no aufs, so we need to port one aufs patch + one fix patch. I've already sent that as a series: [SRU][J][hwe-5.17][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files will resent with clear branches indication ============================ * jammy/linux - master / master-next aufs is present, all that we need to remove excess IS_ENABLED(CONFIG_AUFS_FS) check in fs/overlayfs/file.c. I've sent this change as: [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency ============================ * kinetic/linux - master * focal/linux hwe-5.15-next hwe-5.13-next aufs is present, but we have no fix commit here at all. Will send. ============================ Regards, Alex > > -Stefan > > > > > Acked-by: Andrea Righi <andrea.righi@canonical.com> > > > > The other kernels that require this fix are: > > > > kinetic/linux: > > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/kinetic > > > > jammy/linux: > > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy > > > > If the fix is applied to both these kernels all the derivatives will > > inherit the fix automatically (there's also hwe-5.17, but you already > > provided a fix for that). > > > > Thanks, > > -Andrea >
On Wed, 3 Aug 2022 14:43:43 +0300 Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com> wrote: > On Wed, 3 Aug 2022 09:24:17 +0200 > Stefan Bader <stefan.bader@canonical.com> wrote: > > > On 03.08.22 08:04, Andrea Righi wrote: > > > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > > > ... > > >>> > > >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > > >>> rush to have the focal patch, I was just pointing out that we need to > > >>> fix also this kernel to make everything aligned/consistent. > > >> > > >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > > > > > Alright, after all the clarifications I change my NACK to an ACK: > > > > Sorry with the amount of things incoming this does not fly well. It it better to > > resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here > > are the target kernel names and if that is omitted it is the primary "linux" source. > > jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is > > the parent of all 5.17 based kernels). Any other HWE kernels normally are > > derived from a primary kernel. So they inherit the changes automatically. > > Dear friends, > > I've rechecked different kernels and this is full summary: > > ============================ > > * jammy/linux > - hwe-5.17 > * kinetic/linux > - master-next > > these branches has no aufs, so we need to port one aufs patch + one fix patch. > I've already sent that as a series: > [SRU][J][hwe-5.17][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > will resent with clear branches indication > > ============================ > > * jammy/linux > - master / master-next > > aufs is present, all that we need to remove excess IS_ENABLED(CONFIG_AUFS_FS) check > in fs/overlayfs/file.c. I've sent this change as: > [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > > ============================ > > * kinetic/linux > - master > * focal/linux > hwe-5.15-next > hwe-5.13-next > > aufs is present, but we have no fix commit here at all. Will send. > > ============================ Have done: [SRU v2][jammy:hwe-5.17][kinetic:master-next][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files [SRU][kinetic:master][focal:hwe-5.13-next/hwe-5.15-next][PATCH] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files For Jammy/master we still need: [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency because it's fixup. But for [kinetic:master][focal:hwe-5.13-next/hwe-5.15-next] I've just combined all fixups into the one fix commit just to make maintainers work easier. Regards, Alex > > Regards, > Alex > > > > -Stefan > > > > > > > > Acked-by: Andrea Righi <andrea.righi@canonical.com> > > > > > > The other kernels that require this fix are: > > > > > > kinetic/linux: > > > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/kinetic > > > > > > jammy/linux: > > > - git://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy > > > > > > If the fix is applied to both these kernels all the derivatives will > > > inherit the fix automatically (there's also hwe-5.17, but you already > > > provided a fix for that). > > > > > > Thanks, > > > -Andrea > > > >
On Wed, Aug 03, 2022 at 02:43:43PM +0300, Alexander Mikhalitsyn wrote: > On Wed, 3 Aug 2022 09:24:17 +0200 > Stefan Bader <stefan.bader@canonical.com> wrote: > > > On 03.08.22 08:04, Andrea Righi wrote: > > > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > > > ... > > >>> > > >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > > >>> rush to have the focal patch, I was just pointing out that we need to > > >>> fix also this kernel to make everything aligned/consistent. > > >> > > >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > > > > > Alright, after all the clarifications I change my NACK to an ACK: > > > > Sorry with the amount of things incoming this does not fly well. It it better to > > resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here > > are the target kernel names and if that is omitted it is the primary "linux" source. > > jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is > > the parent of all 5.17 based kernels). Any other HWE kernels normally are > > derived from a primary kernel. So they inherit the changes automatically. > > Dear friends, > > I've rechecked different kernels and this is full summary: > > ============================ > > * jammy/linux > - hwe-5.17 > * kinetic/linux > - master-next > > these branches has no aufs, so we need to port one aufs patch + one fix patch. > I've already sent that as a series: > [SRU][J][hwe-5.17][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > will resent with clear branches indication > > ============================ > > * jammy/linux > - master / master-next > > aufs is present, all that we need to remove excess IS_ENABLED(CONFIG_AUFS_FS) check > in fs/overlayfs/file.c. I've sent this change as: > [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > > ============================ > > * kinetic/linux > - master > * focal/linux > hwe-5.15-next > hwe-5.13-next > > aufs is present, but we have no fix commit here at all. Will send. > > ============================ > > Regards, > Alex Thanks Alex, keep in mind that we don't need special fixes for hwe-5.15 or hwe-5.13, because these kernels are derived from other non-hwe kernels (specifically jammy/linux and impish/linux), so they will automatically get the fixes from their parent. Then there's hwe-5.17 that is (unfortunately) a special case, it started as a derived kernel (from linux-unstable 5.17), and now it's detached from linux-unstable, so it's a standalone kernel and it doesn't have a parent anymore. To summarize I think all we need is the fixes for: - kinetic/linux - jammy/linux - jammy/hwe-5.17 Thanks, -Andrea
On Wed, 3 Aug 2022 16:51:01 +0200 Andrea Righi <andrea.righi@canonical.com> wrote: > On Wed, Aug 03, 2022 at 02:43:43PM +0300, Alexander Mikhalitsyn wrote: > > On Wed, 3 Aug 2022 09:24:17 +0200 > > Stefan Bader <stefan.bader@canonical.com> wrote: > > > > > On 03.08.22 08:04, Andrea Righi wrote: > > > > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > > > > ... > > > >>> > > > >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > > > >>> rush to have the focal patch, I was just pointing out that we need to > > > >>> fix also this kernel to make everything aligned/consistent. > > > >> > > > >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > > > > > > > Alright, after all the clarifications I change my NACK to an ACK: > > > > > > Sorry with the amount of things incoming this does not fly well. It it better to > > > resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here > > > are the target kernel names and if that is omitted it is the primary "linux" source. > > > jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is > > > the parent of all 5.17 based kernels). Any other HWE kernels normally are > > > derived from a primary kernel. So they inherit the changes automatically. > > > > Dear friends, > > > > I've rechecked different kernels and this is full summary: > > > > ============================ > > > > * jammy/linux > > - hwe-5.17 > > * kinetic/linux > > - master-next > > > > these branches has no aufs, so we need to port one aufs patch + one fix patch. > > I've already sent that as a series: > > [SRU][J][hwe-5.17][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > > > will resent with clear branches indication > > > > ============================ > > > > * jammy/linux > > - master / master-next > > > > aufs is present, all that we need to remove excess IS_ENABLED(CONFIG_AUFS_FS) check > > in fs/overlayfs/file.c. I've sent this change as: > > [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > > > > ============================ > > > > * kinetic/linux > > - master > > * focal/linux > > hwe-5.15-next > > hwe-5.13-next > > > > aufs is present, but we have no fix commit here at all. Will send. > > > > ============================ > > > > Regards, > > Alex > > Thanks Alex, Hi Andrea, > > keep in mind that we don't need special fixes for hwe-5.15 or hwe-5.13, > because these kernels are derived from other non-hwe kernels > (specifically jammy/linux and impish/linux), so they will automatically > get the fixes from their parent. > > Then there's hwe-5.17 that is (unfortunately) a special case, it started > as a derived kernel (from linux-unstable 5.17), and now it's detached > from linux-unstable, so it's a standalone kernel and it doesn't have a > parent anymore. Thanks for clarifying! > > To summarize I think all we need is the fixes for: > - kinetic/linux [SRU][kinetic:master][focal:hwe-5.13-next/hwe-5.15-next][PATCH] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > - jammy/linux [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > - jammy/hwe-5.17 [SRU v2][jammy:hwe-5.17][kinetic:master-next][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files OK, then I think that's all from my side because I've sent all the patches. Please, let me know if any problems appear. I'm also ready to test and verify proposed experimental kernel builds. Regards, Alex > > Thanks, > -Andrea
On Wed, Aug 03, 2022 at 06:01:07PM +0300, Alexander Mikhalitsyn wrote: > On Wed, 3 Aug 2022 16:51:01 +0200 > Andrea Righi <andrea.righi@canonical.com> wrote: > > > On Wed, Aug 03, 2022 at 02:43:43PM +0300, Alexander Mikhalitsyn wrote: > > > On Wed, 3 Aug 2022 09:24:17 +0200 > > > Stefan Bader <stefan.bader@canonical.com> wrote: > > > > > > > On 03.08.22 08:04, Andrea Righi wrote: > > > > > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > > > > > ... > > > > >>> > > > > >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > > > > >>> rush to have the focal patch, I was just pointing out that we need to > > > > >>> fix also this kernel to make everything aligned/consistent. > > > > >> > > > > >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > > > > > > > > > Alright, after all the clarifications I change my NACK to an ACK: > > > > > > > > Sorry with the amount of things incoming this does not fly well. It it better to > > > > resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here > > > > are the target kernel names and if that is omitted it is the primary "linux" source. > > > > jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is > > > > the parent of all 5.17 based kernels). Any other HWE kernels normally are > > > > derived from a primary kernel. So they inherit the changes automatically. > > > > > > Dear friends, > > > > > > I've rechecked different kernels and this is full summary: > > > > > > ============================ > > > > > > * jammy/linux > > > - hwe-5.17 > > > * kinetic/linux > > > - master-next > > > > > > these branches has no aufs, so we need to port one aufs patch + one fix patch. > > > I've already sent that as a series: > > > [SRU][J][hwe-5.17][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > > > > > will resent with clear branches indication > > > > > > ============================ > > > > > > * jammy/linux > > > - master / master-next > > > > > > aufs is present, all that we need to remove excess IS_ENABLED(CONFIG_AUFS_FS) check > > > in fs/overlayfs/file.c. I've sent this change as: > > > [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > > > > > > ============================ > > > > > > * kinetic/linux > > > - master > > > * focal/linux > > > hwe-5.15-next > > > hwe-5.13-next > > > > > > aufs is present, but we have no fix commit here at all. Will send. > > > > > > ============================ > > > > > > Regards, > > > Alex > > > > Thanks Alex, > > Hi Andrea, > > > > > keep in mind that we don't need special fixes for hwe-5.15 or hwe-5.13, > > because these kernels are derived from other non-hwe kernels > > (specifically jammy/linux and impish/linux), so they will automatically > > get the fixes from their parent. > > > > Then there's hwe-5.17 that is (unfortunately) a special case, it started > > as a derived kernel (from linux-unstable 5.17), and now it's detached > > from linux-unstable, so it's a standalone kernel and it doesn't have a > > parent anymore. > > Thanks for clarifying! > > > > > To summarize I think all we need is the fixes for: > > - kinetic/linux > > [SRU][kinetic:master][focal:hwe-5.13-next/hwe-5.15-next][PATCH] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > > - jammy/linux > > [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > > > - jammy/hwe-5.17 > > [SRU v2][jammy:hwe-5.17][kinetic:master-next][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > OK, then I think that's all from my side because I've sent all the patches. > > Please, let me know if any problems appear. I'm also ready to test and verify proposed experimental kernel builds. Very good thanks! I'll sort out your fixes and submit emails with the proper subject so that everything can be applied to the right kernels (and will keep your signed-off-by line of course). Thank you so much for looking at this! -Andrea
On Wed, 3 Aug 2022 17:11:38 +0200 Andrea Righi <andrea.righi@canonical.com> wrote: > On Wed, Aug 03, 2022 at 06:01:07PM +0300, Alexander Mikhalitsyn wrote: > > On Wed, 3 Aug 2022 16:51:01 +0200 > > Andrea Righi <andrea.righi@canonical.com> wrote: > > > > > On Wed, Aug 03, 2022 at 02:43:43PM +0300, Alexander Mikhalitsyn wrote: > > > > On Wed, 3 Aug 2022 09:24:17 +0200 > > > > Stefan Bader <stefan.bader@canonical.com> wrote: > > > > > > > > > On 03.08.22 08:04, Andrea Righi wrote: > > > > > > On Tue, Aug 02, 2022 at 04:35:36PM +0300, Alexander Mikhalitsyn wrote: > > > > > > ... > > > > > >>> > > > > > >>> Yes, I mean focal/hwe-5.15 (that is derived from jammy/linux) and ok, no > > > > > >>> rush to have the focal patch, I was just pointing out that we need to > > > > > >>> fix also this kernel to make everything aligned/consistent. > > > > > >> > > > > > >> Sure, I'm ready to prepare patches for all supported kernels. But I will need some guidance from you guys regarding the list of proper branches/git repositories :) > > > > > > > > > > > > Alright, after all the clarifications I change my NACK to an ACK: > > > > > > > > > > Sorry with the amount of things incoming this does not fly well. It it better to > > > > > resubmit as "[SRU K/J v2] ...". The branch is not necessary. More important here > > > > > are the target kernel names and if that is omitted it is the primary "linux" source. > > > > > jammy:linux-hwe-5.17 is only needed separately because it has no parent (but is > > > > > the parent of all 5.17 based kernels). Any other HWE kernels normally are > > > > > derived from a primary kernel. So they inherit the changes automatically. > > > > > > > > Dear friends, > > > > > > > > I've rechecked different kernels and this is full summary: > > > > > > > > ============================ > > > > > > > > * jammy/linux > > > > - hwe-5.17 > > > > * kinetic/linux > > > > - master-next > > > > > > > > these branches has no aufs, so we need to port one aufs patch + one fix patch. > > > > I've already sent that as a series: > > > > [SRU][J][hwe-5.17][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > > > > > > > will resent with clear branches indication > > > > > > > > ============================ > > > > > > > > * jammy/linux > > > > - master / master-next > > > > > > > > aufs is present, all that we need to remove excess IS_ENABLED(CONFIG_AUFS_FS) check > > > > in fs/overlayfs/file.c. I've sent this change as: > > > > [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > > > > > > > > ============================ > > > > > > > > * kinetic/linux > > > > - master > > > > * focal/linux > > > > hwe-5.15-next > > > > hwe-5.13-next > > > > > > > > aufs is present, but we have no fix commit here at all. Will send. > > > > > > > > ============================ > > > > > > > > Regards, > > > > Alex > > > > > > Thanks Alex, > > > > Hi Andrea, > > > > > > > > keep in mind that we don't need special fixes for hwe-5.15 or hwe-5.13, > > > because these kernels are derived from other non-hwe kernels > > > (specifically jammy/linux and impish/linux), so they will automatically > > > get the fixes from their parent. > > > > > > Then there's hwe-5.17 that is (unfortunately) a special case, it started > > > as a derived kernel (from linux-unstable 5.17), and now it's detached > > > from linux-unstable, so it's a standalone kernel and it doesn't have a > > > parent anymore. > > > > Thanks for clarifying! > > > > > > > > To summarize I think all we need is the fixes for: > > > - kinetic/linux > > > > [SRU][kinetic:master][focal:hwe-5.13-next/hwe-5.15-next][PATCH] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > > > > - jammy/linux > > > > [SRU][J][master-next][PATCH] UBUNTU: SAUCE: overlayfs: remove CONFIG_AUFS_FS dependency > > > > > - jammy/hwe-5.17 > > > > [SRU v2][jammy:hwe-5.17][kinetic:master-next][PATCH 0/2] UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files > > > > OK, then I think that's all from my side because I've sent all the patches. > > > > Please, let me know if any problems appear. I'm also ready to test and verify proposed experimental kernel builds. > > Very good thanks! I'll sort out your fixes and submit emails with the > proper subject so that everything can be applied to the right kernels > (and will keep your signed-off-by line of course). Sure, thanks! > > Thank you so much for looking at this! Not at all. Thank you guys for helping me with this and it's really important for CRIU project :) Regards, Alex > > -Andrea
diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c index 2e4ebebdb7d1..7024771dcae8 100644 --- a/fs/overlayfs/file.c +++ b/fs/overlayfs/file.c @@ -490,7 +490,6 @@ static int ovl_fsync(struct file *file, loff_t start, loff_t end, int datasync) return ret; } -#if IS_ENABLED(CONFIG_AUFS_FS) /* * In map_files_get_link() (fs/proc/base.c) * we need to determine correct path from overlayfs. @@ -516,12 +515,6 @@ static void ovl_vm_prfile_set(struct vm_area_struct *vma, vma->vm_region->vm_prfile = file; #endif } -#else /* !CONFIG_AUFS_FS */ -static void ovl_vm_prfile_set(struct vm_area_struct *vma, - struct file *file) -{ -} -#endif/* CONFIG_AUFS_FS */ static int ovl_mmap(struct file *file, struct vm_area_struct *vma) {
BugLink: https://bugs.launchpad.net/bugs/1967924 BugLink: https://bugs.launchpad.net/bugs/1857257 BugLink: https://bugs.launchpad.net/bugs/1967924 Right now we have a fix: b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") in master branch of Jammy kernel, but only formaly. Because these kernels compiled without CONFIG_AUFS_FS set, so this fix just disabled. There is no need to make it dependent on CONFIG_AUFS_FS option, because in all cases we have mm/prfile.c compiled-in. Cc: Andrei Vagin <avagin@gmail.com> Cc: Adrian Reber <areber@redhat.com> Cc: Stefan Bader <stefan.bader@canonical.com> Cc: Kleber Sacilotto de Souza <kleber.souza@canonical.com> Cc: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com> Cc: Andrea Righi <andrea.righi@canonical.com> Fixes: b07bc17b ("UBUNTU: SAUCE: overlayfs: fix incorrect mnt_id of files opened from map_files") Signed-off-by: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com> --- fs/overlayfs/file.c | 7 ------- 1 file changed, 7 deletions(-)