[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS #99699

tpetazzoni · 2024-07-19T20:26:44Z

This change is identical to
26800a2 ("[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS on Linux"), but for sanitizer_procmaps_solaris.cpp.

Indeed, even though sanitizer_procmaps_solaris.cpp is Solaris specific, it also gets built on Linux platforms. It also includes sanitizer_platform.h, which also ends up including features-time64.h, causing a build failure on 32-bit Linux platforms on which 64-bit time_t is enabled by setting _TIME_BITS=64.

To fix this, we do the same change: undefine _TIME_BITS, which anyway will cause no harm as the rest of this file is inside a SANITIZER_SOLARIS compile-time conditional.

Fixes:

In file included from /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features.h:394,
from ../../../../libsanitizer/sanitizer_common/sanitizer_platform.h:25,
from ../../../../libsanitizer/sanitizer_common/sanitizer_procmaps_solaris.cpp:14:
/home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features-time64.h:26:5: error: #error "_TIME_BITS=64 is al lowed only with _FILE_OFFSET_BITS=64"
26 | # error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64"
| ^~~~~

github-actions · 2024-07-19T20:27:01Z

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be
notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write
permissions for the repository. In which case you can instead tag reviewers by
name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review
by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate
is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

llvmbot · 2024-07-19T20:27:28Z

@llvm/pr-subscribers-compiler-rt-sanitizer

Author: Thomas Petazzoni (tpetazzoni)

Changes

This change is identical to
26800a2 ("[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS on Linux"), but for sanitizer_procmaps_solaris.cpp.

Indeed, even though sanitizer_procmaps_solaris.cpp is Solaris specific, it also gets built on Linux platforms. It also includes sanitizer_platform.h, which also ends up including features-time64.h, causing a build failure on 32-bit Linux platforms on which 64-bit time_t is enabled by setting _TIME_BITS=64.

To fix this, we do the same change: undefine _TIME_BITS, which anyway will cause no harm as the rest of this file is inside a SANITIZER_SOLARIS compile-time conditional.

Fixes:

In file included from /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features.h:394,
from ../../../../libsanitizer/sanitizer_common/sanitizer_platform.h:25,
from ../../../../libsanitizer/sanitizer_common/sanitizer_procmaps_solaris.cpp:14:
/home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features-time64.h:26:5: error: #error "_TIME_BITS=64 is al lowed only with _FILE_OFFSET_BITS=64"
26 | # error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64"
| ^~~~~

Full diff: https://github.com/llvm/llvm-project/pull/99699.diff

1 Files Affected:

(modified) compiler-rt/lib/sanitizer_common/sanitizer_procmaps_solaris.cpp (+1)

diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_procmaps_solaris.cpp b/compiler-rt/lib/sanitizer_common/sanitizer_procmaps_solaris.cpp
index eeb49e2afe34d..1b23fd4d512b8 100644
--- a/compiler-rt/lib/sanitizer_common/sanitizer_procmaps_solaris.cpp
+++ b/compiler-rt/lib/sanitizer_common/sanitizer_procmaps_solaris.cpp
@@ -11,6 +11,7 @@
 
 // Before Solaris 11.4, <procfs.h> doesn't work in a largefile environment.
 #undef _FILE_OFFSET_BITS
+#undef _TIME_BITS
 #include "sanitizer_platform.h"
 #if SANITIZER_SOLARIS
 #  include <fcntl.h>

rorth · 2024-07-19T21:54:46Z

Thanks for the patch. A couple of comments:

You should request one or more reviewers (the last ones to review the file(s) in question are a good bet), otherwise the patch is most likely overlooked/ignored. I only noticed it due to your equivalent gcc-patches post.
Please indicate which target (triple) you noticed the problem on and where/how you tested the patch.
The #undef needs a comment explaining why it's needed, especially given that this is a Solaris-specific file and Solaris doesn't have/care about _TIME_BITS.

tpetazzoni · 2024-07-20T09:27:33Z

Thanks for the patch. A couple of comments:

Thanks a lot for the quick feedback!

* You should request one or more reviewers (the last ones to review the file(s) in question are a good bet), otherwise the patch is most likely overlooked/ignored.  I only noticed it due to your equivalent `gcc-patches` post.

Thanks for noticing my patch then. It looks like you are the last person who made changes to this file, so I guess you're the right person to request a review from. I see @richlowe also contributed to this file 3 years ago.

* Please indicate which target (triple) you noticed the problem on and where/how you tested the patch.

The patch was actually tested on the GCC code base, when building an i686 glibc toolchain. But the problem exists for all 32-bit architectures for which libsanitizer has support for, when building with -D_TIME_BITS=64.

* The `#undef` needs a comment explaining why it's needed, especially given that this is a Solaris-specific file and Solaris doesn't have/care about `_TIME_BITS`.

It may be a Solaris-specific file, but it gets compiled unconditionally regardless of the current platform. I'll add a comment that explains things.

tpetazzoni · 2024-07-20T09:30:14Z

I have force pushed a new version of the patch, with a more detailed comment.

mgorny

I can confirm that it fixes the build failure on a time64 x86 system, and given that I wanted to make an identical change myself, LGTM.

mgorny · 2024-10-07T17:46:18Z

Also, I think we should backport this to 19.x too.

devnexen

looks reasonable, but @rorth is the solaris expert.

rorth

Although I appreciate having a comment here, this seems way too long to me. Maybe something like

Avoid conflict between `_TIME_BITS` defined vs. `_FILE_OFFSET_BITS` undefined in some Linux configurations.

or something like this. Certainly not more than a line or two.

This won't affect Solaris at all: its headers contain no reference to _TIME_BITS whatsoever.

mgorny · 2024-10-10T13:15:10Z

@tpetazzoni, sorry to ping you, but are you still around?

This change is identical to 26800a2 ("[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS on Linux"), but for sanitizer_procmaps_solaris.cpp. Indeed, even though sanitizer_procmaps_solaris.cpp is Solaris specific, it also gets built on Linux platforms. It also includes sanitizer_platform.h, which also ends up including features-time64.h, causing a build failure on 32-bit Linux platforms on which 64-bit time_t is enabled by setting _TIME_BITS=64. To fix this, we do the same change: undefine _TIME_BITS, which anyway will cause no harm as the rest of this file is inside a SANITIZER_SOLARIS compile-time conditional. Fixes: In file included from /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features.h:394, from ../../../../libsanitizer/sanitizer_common/sanitizer_platform.h:25, from ../../../../libsanitizer/sanitizer_common/sanitizer_procmaps_solaris.cpp:14: /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features-time64.h:26:5: error: #error "_TIME_BITS=64 is al lowed only with _FILE_OFFSET_BITS=64" 26 | # error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64" | ^~~~~ Signed-off-by: Thomas Petazzoni <[email protected]>

tpetazzoni · 2024-10-10T14:39:48Z

Sorry for the delay, and thanks for the ping. I just pushed a new update, with the commit rebased on the current main, and a shorter comment, as requested by @rorth. Thanks for the review!

rorth

LGTM. Thanks for your patience.

mgorny · 2024-10-14T11:55:15Z

@tpetazzoni, can you merge it or do you need us to merge it for you?

thesamesam · 2024-10-14T19:02:42Z

Pushed, thank you!

This change is identical to 26800a2 ("[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS on Linux"), but for sanitizer_procmaps_solaris.cpp. Indeed, even though sanitizer_procmaps_solaris.cpp is Solaris specific, it also gets built on Linux platforms. It also includes sanitizer_platform.h, which also ends up including features-time64.h, causing a build failure on 32-bit Linux platforms on which 64-bit time_t is enabled by setting _TIME_BITS=64. To fix this, we do the same change: undefine _TIME_BITS, which anyway will cause no harm as the rest of this file is inside a SANITIZER_SOLARIS compile-time conditional. Fixes: In file included from /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features.h:394, from ../../../../libsanitizer/sanitizer_common/sanitizer_platform.h:25, from ../../../../libsanitizer/sanitizer_common/sanitizer_procmaps_solaris.cpp:14: /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features-time64.h:26:5: error: #error "_TIME_BITS=64 is al lowed only with _FILE_OFFSET_BITS=64" 26 | # error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64" | ^~~~~ Signed-off-by: Thomas Petazzoni <[email protected]> Closes: llvm#99699 (cherry picked from commit a121702)

This change is identical to 26800a2 ("[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS on Linux"), but for sanitizer_procmaps_solaris.cpp. Indeed, even though sanitizer_procmaps_solaris.cpp is Solaris specific, it also gets built on Linux platforms. It also includes sanitizer_platform.h, which also ends up including features-time64.h, causing a build failure on 32-bit Linux platforms on which 64-bit time_t is enabled by setting _TIME_BITS=64. To fix this, we do the same change: undefine _TIME_BITS, which anyway will cause no harm as the rest of this file is inside a SANITIZER_SOLARIS compile-time conditional. Fixes: In file included from /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features.h:394, from ../../../../libsanitizer/sanitizer_common/sanitizer_platform.h:25, from ../../../../libsanitizer/sanitizer_common/sanitizer_procmaps_solaris.cpp:14: /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features-time64.h:26:5: error: #error "_TIME_BITS=64 is al lowed only with _FILE_OFFSET_BITS=64" 26 | # error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64" | ^~~~~ Signed-off-by: Thomas Petazzoni <[email protected]> Closes: llvm#99699

This change is identical to 26800a2 ("[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS on Linux"), but for sanitizer_procmaps_solaris.cpp. Indeed, even though sanitizer_procmaps_solaris.cpp is Solaris specific, it also gets built on Linux platforms. It also includes sanitizer_platform.h, which also ends up including features-time64.h, causing a build failure on 32-bit Linux platforms on which 64-bit time_t is enabled by setting _TIME_BITS=64. To fix this, we do the same change: undefine _TIME_BITS, which anyway will cause no harm as the rest of this file is inside a SANITIZER_SOLARIS compile-time conditional. Fixes: In file included from /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features.h:394, from ../../../../libsanitizer/sanitizer_common/sanitizer_platform.h:25, from ../../../../libsanitizer/sanitizer_common/sanitizer_procmaps_solaris.cpp:14: /home/thomas/buildroot/buildroot/output/host/i686-buildroot-linux-gnu/sysroot/usr/include/features-time64.h:26:5: error: #error "_TIME_BITS=64 is al lowed only with _FILE_OFFSET_BITS=64" 26 | # error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64" | ^~~~~ Signed-off-by: Thomas Petazzoni <[email protected]> Closes: llvm#99699 (cherry picked from commit a121702)

A number of items are removed because the issues have been resolved with recipe patches (in separate commits). Some issues were resolved via upstream version updates that bring in needed fixes: glib-2.0 update to 2.78.0 that includes: https://gitlab.gnome.org/GNOME/glib/-/merge_requests/3547 https://gitlab.gnome.org/GNOME/glib/-/merge_requests/3550 curl update to 8.3.0 that includes curl/curl#11610 util-linux update to 2.39 that includes util-linux/util-linux#2430 util-linux/util-linux@3ab9e69 util-linux/util-linux#2435 glib-networking update to 2.78.0 that includes https://gitlab.gnome.org/GNOME/glib-networking/-/merge_requests/241 python3-cryptography update to 42.0.0 which resolves pyca/cryptography#9370 via pyca/cryptography#9964 perl update to 5.40.0 which includes Perl/perl5#21379 python3 update to 3.13.0 which includes python/cpython#118425 python3 update to 3.13.1 which includes python/cpython#124972 tcl update to 9.0.0 which includes tcltk/tcl@4ca6172 (tcl8 recipe has a simple backport of this) dbus update to 1.16.0 which includes https://gitlab.freedesktop.org/dbus/dbus/-/merge_requests/444 https://gitlab.freedesktop.org/dbus/dbus/-/merge_requests/289 openssh update to 10.0p1 which includes openssh/openssh-portable#425 https://bugzilla.mindrot.org/show_bug.cgi?id=3684 https://marc.info/?l=openbsd-bugs&m=172561736524815&w=2 https://lists.mindrot.org/pipermail/openssh-unix-dev/2024-October/041621.html (all reporting the same issue) gcc update to 15.1 which includes llvm/llvm-project#99699 via gcc-mirror/gcc@fa32100 and allows dropping special flags and exceptions for gcc-sanitizers. Signed-off-by: Alexander Kanavin <[email protected]>

llvmbot added compiler-rt compiler-rt:sanitizer labels Jul 19, 2024

tpetazzoni force-pushed the fix-sanitizer-64bit-time branch from 716a024 to 920de81 Compare July 20, 2024 09:29

mgorny requested review from MaskRay, rorth and devnexen October 7, 2024 17:44

mgorny approved these changes Oct 7, 2024

View reviewed changes

devnexen approved these changes Oct 7, 2024

View reviewed changes

rorth reviewed Oct 8, 2024

View reviewed changes

tpetazzoni force-pushed the fix-sanitizer-64bit-time branch from 920de81 to 9be1bde Compare October 10, 2024 14:35

rorth approved these changes Oct 11, 2024

View reviewed changes

thesamesam approved these changes Oct 14, 2024

View reviewed changes

thesamesam closed this in a121702 Oct 14, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS #99699

[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS #99699

Uh oh!

tpetazzoni commented Jul 19, 2024

Uh oh!

github-actions bot commented Jul 19, 2024

Uh oh!

llvmbot commented Jul 19, 2024

Uh oh!

rorth commented Jul 19, 2024

Uh oh!

tpetazzoni commented Jul 20, 2024

Uh oh!

tpetazzoni commented Jul 20, 2024

Uh oh!

mgorny left a comment

Uh oh!

mgorny commented Oct 7, 2024

Uh oh!

devnexen left a comment

Uh oh!

rorth left a comment

Uh oh!

mgorny commented Oct 10, 2024

Uh oh!

tpetazzoni commented Oct 10, 2024

Uh oh!

rorth left a comment

Uh oh!

mgorny commented Oct 14, 2024

Uh oh!

thesamesam commented Oct 14, 2024

Uh oh!

Uh oh!

[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS #99699

[sanitizer] Undef _TIME_BITS along with _FILE_OFFSET_BITS #99699

Uh oh!

Conversation

tpetazzoni commented Jul 19, 2024

Uh oh!

github-actions bot commented Jul 19, 2024

Uh oh!

llvmbot commented Jul 19, 2024

Uh oh!

rorth commented Jul 19, 2024

Uh oh!

tpetazzoni commented Jul 20, 2024

Uh oh!

tpetazzoni commented Jul 20, 2024

Uh oh!

mgorny left a comment

Choose a reason for hiding this comment

Uh oh!

mgorny commented Oct 7, 2024

Uh oh!

devnexen left a comment

Choose a reason for hiding this comment

Uh oh!

rorth left a comment

Choose a reason for hiding this comment

Uh oh!

mgorny commented Oct 10, 2024

Uh oh!

tpetazzoni commented Oct 10, 2024

Uh oh!

rorth left a comment

Choose a reason for hiding this comment

Uh oh!

mgorny commented Oct 14, 2024

Uh oh!

thesamesam commented Oct 14, 2024

Uh oh!

Uh oh!