Mir

Merge lp:~alan-griffiths/mir/fix-1208774 into lp:~mir-team/mir/trunk

Proposed by Alan Griffiths
Status: Merged
Approved by: Robert Ancell
Approved revision: no longer in the source branch.
Merged at revision: 959
Proposed branch: lp:~alan-griffiths/mir/fix-1208774
Merge into: lp:~mir-team/mir/trunk
Diff against target: 10 lines (+4/-0)
1 file modified
src/platform/graphics/CMakeLists.txt (+4/-0)
To merge this branch: bzr merge lp:~alan-griffiths/mir/fix-1208774
Reviewer Review Type Date Requested Status
Robert Ancell Approve
PS Jenkins bot (community) continuous-integration Approve
Alexandros Frantzis (community) Approve
Daniel van Vugt Abstain
Review via email: mp+178925@code.launchpad.net

Commit message

platform: workaround link errors on i386/g++ 4.8.1

Description of the change

platform: workaround link errors on i386/g++ 4.8.1

To post a comment you must log in.
Revision history for this message
Daniel van Vugt (vanvugt) wrote :

Do we have evidence the error is spurious? The tools are usually quite accurate on this. I suspect we have made a serious linkage mistake...

review: Needs Information
Revision history for this message
Alan Griffiths (alan-griffiths) wrote :

> Do we have evidence the error is spurious? The tools are usually quite
> accurate on this. I suspect we have made a serious linkage mistake...

I'm sure the linker is reporting what it sees accurately, but there is no problem with either 64bit or clang builds - which I'd expect to fail the same way if we'd made a serious mistake.

I can't spot anything bad and the "missing" symbol is resolved correctly at runtime. Consequently, I don't think it worth spending more time on.

Revision history for this message
PS Jenkins bot (ps-jenkins) wrote :
review: Approve (continuous-integration)
Revision history for this message
Daniel van Vugt (vanvugt) :
review: Abstain
Revision history for this message
Alexandros Frantzis (afrantzis) wrote :

OK as a workaround. Alternatively we could explicitly link with '-lc' which also resolves the problem (still a workaround, since the linker is linking with -lc anyway).

review: Approve
Revision history for this message
PS Jenkins bot (ps-jenkins) wrote :
review: Needs Fixing (continuous-integration)
Revision history for this message
Robert Ancell (robert-ancell) wrote :

Needs merge..

Revision history for this message
Daniel van Vugt (vanvugt) wrote :

If '-lc' is a viable workaround then that does suggest we're mixing static libc with shared libc. That's a high-risk mistake that can easily lead to heap corruption in the least.

Revision history for this message
PS Jenkins bot (ps-jenkins) wrote :
review: Approve (continuous-integration)
Revision history for this message
Robert Ancell (robert-ancell) :
review: Approve

Preview Diff

[H/L] Next/Prev Comment, [J/K] Next/Prev File, [N/P] Next/Prev Hunk
1=== modified file 'src/platform/graphics/CMakeLists.txt'
2--- src/platform/graphics/CMakeLists.txt 2013-08-08 02:34:31 +0000
3+++ src/platform/graphics/CMakeLists.txt 2013-08-09 08:26:25 +0000
4@@ -1,3 +1,7 @@
5+# Workaround for bug 1208774 (spurious link errors on 32bit g++)
6+string (REPLACE " -Wl,--no-undefined" " " CMAKE_SHARED_LINKER_FLAGS ${CMAKE_SHARED_LINKER_FLAGS})
7+include_directories(${GLESv2_INCLUDE_DIRS})
8+
9 set(
10 GRAPHICS_SOURCES
11

Subscribers

People subscribed via source and target branches