1555530 - [s390x] build errors: invalid operand for instruction; invalid use of length addressing

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Description

[Charles Robertson]

Attachment: Firefox 68 ESR beta build log

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Firefox/60.0

Steps to reproduce:

Build Firefox 68 ESR (beta) on s390x.

Actual results:

Build/compile errors (see attached build log file):
...
[ 553s] 6:08.70 In file included from /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/md/unix/xptcstubs_linux_s390x.cpp:194:
[ 553s] 6:08.70 /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/xptcstubsdef.inc:2:1: error: invalid use of length addressing
[ 553s] 6:08.70 STUB_ENTRY(4)
[ 553s] 6:08.70 ^
[ 553s] 6:08.70 /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/md/unix/xptcstubs_linux_s390x.cpp:167:9: note: expanded from macro 'STUB_ENTRY'
[ 553s] 6:08.70 "lg %0,0(15)\n\t"
[ 553s] 6:08.70 ^
[ 553s] 6:08.70 <inline asm>:1:13: note: instantiated into assembly here
[ 553s] 6:08.70 lg %r1,0(15)
[ 553s] 6:08.70 ^
[ 553s] 6:08.70 In file included from /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/md/unix/xptcstubs_linux_s390x.cpp:194:
[ 553s] 6:08.70 /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/xptcstubsdef.inc:2:1: error: invalid operand for instruction
[ 553s] 6:08.70 STUB_ENTRY(4)
[ 553s] 6:08.70 ^
[ 553s] 6:08.70 /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/md/unix/xptcstubs_linux_s390x.cpp:168:25: note: expanded from macro 'STUB_ENTRY'
[ 553s] 6:08.70 "aghi %0,160\n\t"
[ 553s] 6:08.70 ^
[ 553s] 6:08.70 <inline asm>:3:9: note: instantiated into assembly here
[ 553s] 6:08.70 stmg 3,6,0(%r14)
[ 553s] 6:08.70 ^
[ 553s] 6:08.70 In file included from /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/md/unix/xptcstubs_linux_s390x.cpp:194:
[ 553s] 6:08.70 /home/abuild/rpmbuild/BUILD/firefox-68.0/xpcom/reflect/xptcall/xptcstubsdef.inc:2:1: error: invalid operand for instruction
[ 553s] 6:08.70 STUB_ENTRY(4)
...

Expected results:

Clean build of Firefox

Comment 1

[msirringhaus]

Still present in firefox-68.0b6

Comment 2

[Rares Doghi, Desktop QA]

Hi, I'll set the component for this issue, but I'm not entirely sure its the correct one. Please change the component to a more suitable one in case its not correct.

Comment 3

[Mike Hommey [:glandium]]

Attachment: Bug 1555530 - Don't use integrated assembler with clang on ppc64 and s390x.

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/11427e893b96
Don't use integrated assembler with clang on ppc64 and s390x. r=froydnj

Comment 5

[Mike Hommey [:glandium]]

Comment on attachment 9069818 [details]
Bug 1555530 - Don't use integrated assembler with clang on ppc64 and s390x.

Beta/Release Uplift Approval Request

• User impact if declined: Build failure on powerpc64el and s390x (tier 3 platforms) when compiling with clang. Would be nice to have for downstreams on ESR68.
• Is this code covered by automated tests?: No
• Has the fix been verified in Nightly?: Yes
• Needs manual test from QE?: No
• If yes, steps to reproduce:
• List of other uplifts needed: None
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): NPOTB for what Mozilla ships.
• String changes made/needed:

Comment 6

[Alexandru Michis [:malexandru]]

https://hg.mozilla.org/mozilla-central/rev/11427e893b96

Comment 7

[Julien Cristau [:jcristau]]

Comment on attachment 9069818 [details]
Bug 1555530 - Don't use integrated assembler with clang on ppc64 and s390x.

tier3 build fix, approved for 68.0b8

Comment 8

[Oana Pop-Rus]

https://hg.mozilla.org/releases/mozilla-beta/rev/aa64aae1b32d