Bug 426181 - 4.1.2 20070925 (Red Hat 4.1.2-33) -fstrict-aliasing causes skipped code
4.1.2 20070925 (Red Hat 4.1.2-33) -fstrict-aliasing causes skipped code
Status: CLOSED WONTFIX
Product: Fedora
Classification: Fedora
Component: gcc (Show other bugs)
8
x86_64 Linux
low Severity high
: ---
: ---
Assigned To: Jakub Jelinek
Fedora Extras Quality Assurance
:
Depends On:
Blocks:
  Show dependency treegraph
 
Reported: 2007-12-18 20:46 EST by Ehud Gavron
Modified: 2007-12-19 02:55 EST (History)
0 users

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Environment:
Last Closed: 2007-12-19 02:55:00 EST
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)


External Trackers
Tracker ID Priority Status Summary Last Updated
GNU Compiler Collection 32328 None None None Never

  None (edit)
Description Ehud Gavron 2007-12-18 20:46:58 EST
Description of problem:

This problem is described in detail in 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32328

Which says gnu.org's 4.1.2 is clean of this bug.

That fix needs to be backported to Redhat 4.1.2-33 and Fedora 4.1.2 20070925

Version-Release number of selected component (if applicable):
Fedora 4.1.2 20070925 (Redhat 4.1.2-33)


How reproducible:


Steps to Reproduce:
1.
2.
3.
  
Actual results:


Expected results:


Additional info:
Comment 1 Ehud Gavron 2007-12-19 00:52:51 EST
The symptoms appear when compiling kernel module 
drivers/net/wireless/b43/rfkill.c

The code that calls KEY_WLAN 1 and KEY_WLAN 0 is optimized out.

This is why Fedora 2.6.22-1.42 (f8 x86_64) worked in lighting the LED and
Fedora 2.6.22-1.49 (same arch) didn't.  Same code, different compilation result.

There are additional examples in the references bugzillas... just thought I'd
add a Fedora-specific one.  I'm not a compiler-internals guy.  Just pointing out
this was patched in the gnu 4.1.2 but not in the RH/F 4.1.2.
Comment 2 Jakub Jelinek 2007-12-19 02:55:00 EST
Kernel modules should be compiled with -fno-strict-aliasing anyway, kernel
violates aliasing in hundreds of places.
If you read the PR properly, it would be clear this is a WONTFIX for 4.1.x/4.2.x,
the changes aren't really backportable to earlier GCC versions and are massive.

Note You need to log in before you can comment on or make changes to this bug.