A couple of Linux 5.16 fixes #12975

rincebrain · 2022-01-14T15:59:58Z

(Draft b/c my first run through the CI failed in the ENOSPC checks for some reason, and because I feel like sendfile shouldn't have broken even in that case unless we want to make that a fatal check error?)

Motivation and Context

I went down the hole for #12971 and adjusted a few things before I found one that helped.

Description

Three patches.

add_disk gained an error return code, so detect and conditionally set error to it
kvmalloc is no longer in mm.h, it's in slab.h
iov_iter_fault_in_readable is now fault_in_iov_iter_readable, so again, detect and map that

How Has This Been Tested?

Ran it through -r sanity and -T functional on a Fedora rawhide VM running the 5.16 kernel

Types of changes

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Performance enhancement (non-breaking change which improves efficiency)
Code cleanup (non-breaking change which makes code smaller or more readable)
Breaking change (fix or feature that would cause existing functionality to change)
Library ABI change (libzfs, libzfs_core, libnvpair, libuutil and libzfsbootenv)
Documentation (a change to man pages or other documentation)

Checklist:

My code follows the OpenZFS code style requirements.
I have updated the documentation accordingly.
I have read the contributing document.
I have added tests to cover my changes.
I have run the ZFS Test Suite with this change applied.
All commit messages are properly formatted and contain Signed-off-by.

module/os/linux/zfs/zvol_os.c

behlendorf

This all looks pretty straight forward, thanks for checking it out.

config/kernel-add-disk.m4

config/kernel-vfs-iov_iter.m4

jonathonf · 2022-01-17T01:37:23Z

As a fairly random data point, this also fixed an issue with profile-sync-daemon using overlayfs over ZFS (working with 5.15, broken with 5.16).

behlendorf · 2022-01-18T22:08:33Z

@rincebrain I believe the draft status could be removed from this PR. While not a comprehensive set of patches for 5.16, these all look good to me and should be included. Unless you have a few more in the works? For example, it appears some work will need to be done to address #12926.

rincebrain · 2022-01-19T01:38:18Z

I can take a look at that, but no, I didn't have any more patches to push in this.

behlendorf

@ckane mind having a look at this.

ckane · 2022-01-20T02:52:02Z

Thanks, I'll build this tonight and test it out on my laptop and give feedback tomorrow. Looks fairly clear to me though.

ckane · 2022-01-22T00:00:09Z

ok seems to be working stable on my laptop

behlendorf · 2022-01-22T00:37:54Z

@rincebrain would you mind squashing these commits appropriately. We've also been using the convention of prefixing Linux kernel compatibility fixes with Linux 5.16 compat: in the subject line. It just makes it easier to make sure we're got all the relevant kernel changes when backporting the commits. For example:

Linux 5.16 compat: Added FALLOC_FL_ZERO_RANGE to allowed flags
Linux 5.16 compat: Added mapping for renamed iov_iter_fault_in_readable
Linux 5.16 compat: Added add_disk check for return
Linux 5.16 compat: Include another header so we find kvmalloc

As it says on the tin - the folio work moved a bunch out of mm.h. Signed-off-by: Rich Ercolani <[email protected]>

add_disk went from void to must-check int return. Signed-off-by: Rich Ercolani <[email protected]>

Linux decided to rename this for some reason. At some point, we should probably invert this mapping, but for now... Signed-off-by: Rich Ercolani <[email protected]>

For us, I think it's always just FALLOC_FL_PUNCH_HOLE with a fake mustache on. Signed-off-by: Rich Ercolani <[email protected]>

rincebrain · 2022-01-22T03:02:12Z

There you go. (I didn't tag the FALLOC one as 5.16 compat because it's not, really - it started whining loudly in 5.16 sometimes, but the actual feature was Linux 3.15...)

…

On Fri, Jan 21, 2022 at 7:38 PM Brian Behlendorf ***@***.***> wrote: @rincebrain <https://github.com/rincebrain> would you mind squashing these commits appropriately. We've also been using the convention of prefixing Linux kernel compatibility fixes with Linux 5.16 compat: in the subject line. It just makes it easier to make sure we're got all the relevant kernel changes when backporting the commits. For example: - Linux 5.16 compat: Added FALLOC_FL_ZERO_RANGE to allowed flags - Linux 5.16 compat: Added mapping for renamed iov_iter_fault_in_readable - Linux 5.16 compat: Added add_disk check for return - Linux 5.16 compat: Include another header so we find kvmalloc — Reply to this email directly, view it on GitHub <#12975 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABUI7PSIJ6DQKZZ5HVDCY3UXH367ANCNFSM5L7FFALQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you were mentioned.Message ID: ***@***.***>

add_disk went from void to must-check int return. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes #12975

Linux decided to rename this for some reason. At some point, we should probably invert this mapping, but for now... Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes #12975

For us, I think it's always just FALLOC_FL_PUNCH_HOLE with a fake mustache on. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes #12975

sunilvaltix · 2022-02-02T05:38:50Z

@behlendorf Is this going to be cherry picked into 2.1.x as well?

jonathonf · 2022-02-02T14:12:31Z

Is this going to be cherry picked into 2.1.x as well?

It's very likely to be picked up into the zfs-2.1.3-staging branch.

As it says on the tin - the folio work moved a bunch out of mm.h. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes #12975

add_disk went from void to must-check int return. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes #12975

Linux decided to rename this for some reason. At some point, we should probably invert this mapping, but for now... Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes #12975

For us, I think it's always just FALLOC_FL_PUNCH_HOLE with a fake mustache on. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes #12975

sunilvaltix · 2022-02-07T21:12:50Z

@jonathonf Is there a release criteria for 2.1.3 release? What is blocking the release of 2.1.3 at this point? I see that patches have landed in staging.

jonathonf · 2022-02-07T21:23:21Z

I don't know - I don't control the release cycle, I just try to help here and there.

tonyhutter · 2022-02-07T21:57:31Z

All these commits are included in our upcoming 2.1.3 release: #13063

There's a few others we'd like to include though (like #13059)

As it says on the tin - the folio work moved a bunch out of mm.h. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

add_disk went from void to must-check int return. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

Linux decided to rename this for some reason. At some point, we should probably invert this mapping, but for now... Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

For us, I think it's always just FALLOC_FL_PUNCH_HOLE with a fake mustache on. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

As it says on the tin - the folio work moved a bunch out of mm.h. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

add_disk went from void to must-check int return. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

Linux decided to rename this for some reason. At some point, we should probably invert this mapping, but for now... Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

For us, I think it's always just FALLOC_FL_PUNCH_HOLE with a fake mustache on. Reviewed-by: Brian Behlendorf <[email protected]> Reviewed-by: Coleman Kane <[email protected]> Signed-off-by: Rich Ercolani <[email protected]> Closes openzfs#12975

rincebrain commented Jan 14, 2022

View reviewed changes

module/os/linux/zfs/zvol_os.c Show resolved Hide resolved

behlendorf reviewed Jan 14, 2022

View reviewed changes

config/kernel-add-disk.m4 Show resolved Hide resolved

config/kernel-vfs-iov_iter.m4 Outdated Show resolved Hide resolved

rincebrain force-pushed the zfs-master-516fixes branch from 4a76f03 to a2678ac Compare January 15, 2022 07:03

satmandu mentioned this pull request Jan 17, 2022

loop mounting images broken with kernel 5.16 #12926

Closed

behlendorf added the Status: Code Review Needed Ready for review and testing label Jan 18, 2022

rincebrain marked this pull request as ready for review January 19, 2022 01:38

behlendorf approved these changes Jan 19, 2022

View reviewed changes

gamanakis mentioned this pull request Jan 19, 2022

Linux 5.17 API updates #12989

Closed

13 tasks

behlendorf added Status: Accepted Ready to integrate (reviewed, tested) and removed Status: Code Review Needed Ready for review and testing labels Jan 21, 2022

behlendorf requested a review from tonyhutter January 21, 2022 18:21

rincebrain added 4 commits January 21, 2022 21:55

Linux 5.16 compat: Check slab.h for kvmalloc

7b92691

As it says on the tin - the folio work moved a bunch out of mm.h. Signed-off-by: Rich Ercolani <[email protected]>

Linux 5.16 compat: Added add_disk check for return

1ca38ed

add_disk went from void to must-check int return. Signed-off-by: Rich Ercolani <[email protected]>

Linux 5.16 compat: Added mapping for iov_iter_fault_in_readable

a0b11a8

Linux decided to rename this for some reason. At some point, we should probably invert this mapping, but for now... Signed-off-by: Rich Ercolani <[email protected]>

Add support for FALLOC_FL_ZERO_RANGE

700bb88

For us, I think it's always just FALLOC_FL_PUNCH_HOLE with a fake mustache on. Signed-off-by: Rich Ercolani <[email protected]>

rincebrain force-pushed the zfs-master-516fixes branch from 6b0e37e to 700bb88 Compare January 22, 2022 02:58

satmandu mentioned this pull request Jan 24, 2022

[5.17] PDE_DATA() replaced by pde_data() #13004

Closed

behlendorf closed this in e86f88a Jan 24, 2022

rigred mentioned this pull request Mar 3, 2022

Issue with loop devices in ZFS v2.1.2 and Linux 5.16.x archzfs/archzfs#432

Closed

rincebrain mentioned this pull request May 2, 2022

iov_iter_fault_in_readable is now fault_in_iov_iter_readable #13401

Open

rincebrain mentioned this pull request Sep 20, 2022

sendfile regression in ZFS arch linux 5.16 2.1.2_5.16.arch1.1-1 not zfs-linux-2.1.2_5.15.13.arch1.1-1 #12971

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

A couple of Linux 5.16 fixes #12975

A couple of Linux 5.16 fixes #12975

rincebrain commented Jan 14, 2022

behlendorf left a comment

jonathonf commented Jan 17, 2022

behlendorf commented Jan 18, 2022

rincebrain commented Jan 19, 2022

behlendorf left a comment

ckane commented Jan 20, 2022

ckane commented Jan 22, 2022

behlendorf commented Jan 22, 2022

rincebrain commented Jan 22, 2022 via email

sunilvaltix commented Feb 2, 2022

jonathonf commented Feb 2, 2022

sunilvaltix commented Feb 7, 2022

jonathonf commented Feb 7, 2022

tonyhutter commented Feb 7, 2022

A couple of Linux 5.16 fixes #12975

A couple of Linux 5.16 fixes #12975

Conversation

rincebrain commented Jan 14, 2022

Motivation and Context

Description

How Has This Been Tested?

Types of changes

Checklist:

behlendorf left a comment

Choose a reason for hiding this comment

jonathonf commented Jan 17, 2022

behlendorf commented Jan 18, 2022

rincebrain commented Jan 19, 2022

behlendorf left a comment

Choose a reason for hiding this comment

ckane commented Jan 20, 2022

ckane commented Jan 22, 2022

behlendorf commented Jan 22, 2022

rincebrain commented Jan 22, 2022 via email

sunilvaltix commented Feb 2, 2022

jonathonf commented Feb 2, 2022

sunilvaltix commented Feb 7, 2022

jonathonf commented Feb 7, 2022

tonyhutter commented Feb 7, 2022