1827566 - dom/base/nsTextFragmentGeneric.h:16:70: error: incomplete type ‘xsimd::batch<short int, xsimd::sse2>’ used in nested name specifier

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Mike Hommey [:glandium]]

Similar to bug 1822901, but for i386 Linux after bug 1811184.

/<<PKGBUILDDIR>>/dom/base/nsTextFragmentGeneric.h: In function ‘int32_t mozilla::FirstNon8Bit(const char16_t*, const char16_t*)’:
/<<PKGBUILDDIR>>/dom/base/nsTextFragmentGeneric.h:38:16: error: ‘any’ is not a member of ‘xsimd’
   38 |     if (xsimd::any((vect & vectmask) != 0)) return i;
      |                ^~~
/<<PKGBUILDDIR>>/dom/base/nsTextFragmentGeneric.h: In instantiation of ‘int32_t mozilla::FirstNon8Bit(const char16_t*, const char16_t*) [with Arch = xsimd::sse2; int32_t = int]’:
/<<PKGBUILDDIR>>/dom/base/nsTextFragment.cpp:183:46:   required from here
/<<PKGBUILDDIR>>/dom/base/nsTextFragmentGeneric.h:16:70: error: incomplete type ‘xsimd::batch<short int, xsimd::sse2>’ used in nested name specifier
   16 |   const uint32_t numUnicharsPerVector = xsimd::batch<int16_t, Arch>::size;
      |                                                                      ^~~~
/<<PKGBUILDDIR>>/dom/base/nsTextFragmentGeneric.h:35:31: error: ‘xsimd::batch<short int, xsimd::sse2> vectmask’ has incomplete type
   35 |   xsimd::batch<int16_t, Arch> vectmask(static_cast<int16_t>(shortMask));
      |                               ^~~~~~~~
/<<PKGBUILDDIR>>/dom/base/nsTextFragmentGeneric.h:37:64: error: incomplete type ‘xsimd::batch<short int, xsimd::sse2>’ used in nested name specifier
   37 |     const auto vect = xsimd::batch<int16_t, Arch>::load_aligned(str + i);
      |                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~

Comment 1

[BugBot [:suhaib / :marco]]

The Bugbug bot thinks this bug should belong to the 'Core::DOM: Core & HTML' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 2

[BugBot [:suhaib / :marco]]

Set release status flags based on info from the regressing bug 1811184

Comment 3

[Suhaib Mujahid [:suhaib]]

:kmag could you please set the severity on this to evaluate it for 112?

Comment 4

[Kris Maglione [:kmag]]

We typically don't assign severities to bugs that don't directly affect end users, but I suppose if this is preventing us from shipping to i386 Linux, it qualifies.

That said, if the failure landed and wasn't backed out, it presumably isn't breaking builds on automation, and therefore doesn't affect any of our tier 1 or 2 platforms. Let me know if I'm wrong.

Comment 5

[Suhaib Mujahid [:suhaib]]

(In reply to Kris Maglione [:kmag] from comment #4)

We typically don't assign severities to bugs that don't directly affect end users, but I suppose if this is preventing us from shipping to i386 Linux, it qualifies.

That said, if the failure landed and wasn't backed out, it presumably isn't breaking builds on automation, and therefore doesn't affect any of our tier 1 or 2 platforms. Let me know if I'm wrong.

Thank you! This answers my question regarding 112, so I'm marking this as wontfix for 112.

My understanding is that all defect bugs should have the severity set.

Comment 6

[[:sergesanspaille]]

Mike, can you share the flags used to compile that file? Looks like our headers think we have sse2 but xsimd disagrees.

Comment 7

[Mike Hommey [:glandium]]

It's reproducible with the following mozconfig:

ac_add_options --target=i686
export CXXFLAGS=-no-sse

(You need to apply the patch from https://bugzilla.mozilla.org/show_bug.cgi?id=1729459#c25 too)

The problem is that xsimd does want sse enabled, but the code that uses it is not compiled with SSE_FLAGS (because it's not SSE-specific code), while being behind the ifdef that works for code that may use sse via runtime detection.

Comment 8

[[:sergesanspaille]]

Attachment: Bug 1827566 - Isolate SSE2 requirements to SSE-compiled file r=glandium

Only provide forward declaration to the generic code.

Comment 9

[[:sergesanspaille]]

Attachment: m32.patch

Comment 10

[[:sergesanspaille]]

Note: to test my change, I had to apply the attached patch m32.patch , which involves more than the patch you quoted

Comment 11

[Mike Hommey [:glandium]]

(In reply to [:sergesanspaille] from comment #10)

Note: to test my change, I had to apply the attached patch m32.patch , which involves more than the patch you quoted

Probably because of the CXXFLAGS=-no-sse, which is an unnatural way to reproduce.

Comment 12

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/5055804013a2
Isolate SSE2 requirements to SSE-compiled file r=glandium

Comment 13

[Cristina Horotan [:chorotan]]

Backed out changeset 5055804013a2 (bug 1827566) for causing build bustage at nsTextFragmentGenericFwd.h

Backout: https://hg.mozilla.org/integration/autoland/rev/c0289365efbe4dea3bdf9de486e272051132f650

Failure push: https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&revision=5055804013a2698d10601e6600cedbcc20603fe8

Failure log: https://treeherder.mozilla.org/logviewer?job_id=413553109&repo=autoland&lineNumber=7274

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/db920151d68a
Isolate SSE2 requirements to SSE-compiled file r=glandium

Comment 15

[Cristina Horotan [:chorotan]]

Backed out changeset db920151d68a (bug 1827566) for causing build bustages at nsTextFragmentGenericFwd.h

Backout: https://hg.mozilla.org/integration/autoland/rev/7b8dc255cf898f693e00bb4209b02c15722ac1d1

Failure push: https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&revision=db920151d68a5562d9291c92ccab62b99fa3ee70&selectedTaskRun=Zu2lOI0mRz683LyIhzBDhQ.0

Failure log: https://treeherder.mozilla.org/logviewer?job_id=413588125&repo=autoland&lineNumber=7556

Comment 16

[[:sergesanspaille]]

Sorry Cristina, I relanded a version that passes OSX build a few minutes ago (associated try build: https://treeherder.mozilla.org/jobs?repo=try&revision=28a845ad3d970e9a871157ad3e9146acc0611949)

Comment 17

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/66811b9fe888
Isolate SSE2 requirements to SSE-compiled file r=glandium

Comment 18

[Cristina Horotan [:chorotan]]

https://hg.mozilla.org/mozilla-central/rev/66811b9fe888

Comment 19

[BugBot [:suhaib / :marco]]

The patch landed in nightly and beta is affected.
:sergesanspaille, is this bug important enough to require an uplift?

• If yes, please nominate the patch for beta approval.
• If no, please set status-firefox113 to wontfix.

For more information, please visit auto_nag documentation.