Allow OMPI to compile on OSX with gcc13 #12117
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
bosilca
force-pushed
the
fix/compile_with_gcc13
branch
from
c480a65 to
ebbdfd5
Compare
jsquyres
reviewed
Nov 23, 2023
There was a problem hiding this comment.
Can you cite @eschnett somewhere? He did report these issues.
Per #12110 (comment), I don't actually have problems compiling OMPI with Homebrew-installed gcc 13.2.0. 🤷♂️
But most of these changes look good, anyway -- the only quibble I have is with SSIZE_MAX (noted above).
jsquyres
approved these changes
Nov 27, 2023
jsquyres
reviewed
Nov 27, 2023
bosilca
force-pushed
the
fix/compile_with_gcc13
branch
from
ebbdfd5 to
fc96057
Compare
jsquyres
force-pushed
the
fix/compile_with_gcc13
branch
from
fc96057 to
4a79146
Compare
jsquyres
reviewed
Jan 18, 2024
jsquyres
approved these changes
Jan 18, 2024
|
all the commits are unverified now. let me try to fix this. |
bosilca
force-pushed
the
fix/compile_with_gcc13
branch
from
4a79146 to
533b8d6
Compare
The only way to get IOV_MAX (and friends) is to manually include the GCC specific syslimits.h header (sys/syslimits.h would not work either). Fixes open-mpi#12110 Signed-off-by: George Bosilca <[email protected]>
Manually define it based on the length of the ssize_t type. Fixes open-mpi#12109 Signed-off-by: George Bosilca <[email protected]>
Instead rely on OPAL_PATH_MAX #define. This complements the fix for open-mpi#12108 Signed-off-by: George Bosilca <[email protected]>
We need to discover the existence and then include sys/param.h while using the correctly named MPL defines. Fixes open-mpi#12108 Signed-off-by: George Bosilca <[email protected]>
bosilca
force-pushed
the
fix/compile_with_gcc13
branch
from
cbadfa5 to
9347a55
Compare
Sorry, that was probably just because I pushed last night (i.e., I became the committer on those commits). The NVIDIA CI was failing last night; let's see if it passes today... |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Thanks @eschnett for reporting the 3 issues related to gcc13.
I am not entirely convinced the root cause is on OMPI side. It seems to me that some versions of gcc13.2 fail to define critical POSIX constants/defines (IOV_MAX, PATH_MAX and SSIZE_MAX) as well as their minimal values. Digging a little around it seems gcc13.2 (at least the MacPort variant) fails to use the OSX system headers to get these constants, aka. basically missing an #include_next on limits.h (this #include_next existed in gcc12).
Anyway, until we figure out the real solution I need to be able to compile on my laptop with gcc13.
Fixes #12108
Fixes #12109
Fixes #12110