Bug 605803

Summary: gcc gets an internal compiler error when compiling a kernel module
Product: Red Hat Enterprise Linux 5 Reporter: David Smith <dsmith>
Component: gccAssignee: Jakub Jelinek <jakub>
Status: CLOSED ERRATA QA Contact: qe-baseos-tools-bugs
Severity: medium Docs Contact:
Priority: low    
Version: 5.5CC: mnowak, pmuller
Target Milestone: rc   
Target Release: ---   
Hardware: ppc64   
OS: Linux   
Whiteboard:
Fixed In Version: gcc-4.1.2-49.el5 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2011-01-13 23:58:30 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
source tar file
none
gcc41-rh605803.patch
none
gcc41-rh605803.patch none

Description David Smith 2010-06-18 19:44:16 UTC 
Created attachment 425240 [details]
source tar file

Description of problem:

When a kernel module (generated by systemtap) is compiled, gcc reports an internal compiler error and gets a segmentation fault.

Version-Release number of selected component (if applicable):

gcc-4.1.2-48.el5
kernel-2.6.18-194.el5
kernel-devel-2.6.18-194.el5

How reproducible:

Happens every time.

Steps to Reproduce:
# tar jxf reproducer.tar.bz2
# make -C /lib/modules/2.6.18-194.el5/build M=`pwd`/reproducer ARCH="powerpc" modules V=1
  
Actual results:

# make -C /lib/modules/2.6.18-194.el5/build M=`pwd`/reproducer ARCH="powerpc" modules V=1
make: Entering directory `/usr/src/kernels/2.6.18-194.el5-ppc64'
test -e include/linux/autoconf.h -a -e include/config/auto.conf || (		\
	echo;								\
	echo "  ERROR: Kernel configuration is invalid.";		\
	echo "         include/linux/autoconf.h or include/config/auto.conf are missing.";	\
	echo "         Run 'make oldconfig && make prepare' on kernel src to fix it.";	\
	echo;								\
	/bin/false)
mkdir -p /root/ppc/reproducer/.tmp_versions
rm -f /root/ppc/reproducer/.tmp_versions/*
make -f scripts/Makefile.build obj=/root/ppc/reproducer
  gcc -m64 -Wp,-MD,/root/ppc/reproducer/.stap_16929.o.d  -nostdinc -isystem /usr/lib/gcc/ppc64-redhat-linux/4.1.2/include -D__KERNEL__ -Iinclude  -include include/linux/autoconf.h  -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -Wstrict-prototypes -Wundef -Werror-implicit-function-declaration -fno-delete-null-pointer-checks -fwrapv -Os -msoft-float -pipe -mminimal-toc -mtraceback=none  -mcall-aixdesc -mtune=power4 -mno-altivec -funit-at-a-time -mstring -Wa,-maltivec -fomit-frame-pointer -g  -fno-stack-protector -Wdeclaration-after-statement -Wno-pointer-sign  -Iinclude2/asm/mach-default -include /tmp/stapyoaCsq/stapconf_5c46f822e8a43ee3ca3292dcc51cfc81_427.h -freorder-blocks  -Wno-unused -Werror -I"/usr/local/share/systemtap/runtime"  -DMODULE -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(stap_16929)"  -D"KBUILD_MODNAME=KBUILD_STR(stap_16929)" -c -o /root/ppc/reproducer/.tmp_stap_16929.o /root/ppc/reproducer/stap_16929.c
/root/ppc/reproducer/stap_16929.c: In function ‘enter_profile_probes’:
/root/ppc/reproducer/stap_16929.c:346: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://bugzilla.redhat.com/bugzilla> for instructions.
Preprocessed source stored into /tmp/cchfsYNu.out file, please attach this to your bugreport.
make[1]: *** [/root/ppc/reproducer/stap_16929.o] Error 1
make: *** [_module_/root/ppc/reproducer] Error 2
make: Leaving directory `/usr/src/kernels/2.6.18-194.el5-ppc64'

Expected results:

A compiled module, not a segfault.

Additional info:

The reproducer tar file includes a preprocessed source file, called 'ccw3R6PE.out'.
Comment 1 Jakub Jelinek 2010-06-21 08:59:28 UTC 
This crashes in set_uids_in_ptset, in
          else if (var_can_have_subvars (vi->decl)
                   && get_subvars_for_var (vi->decl))
            {
              /* If VI->DECL is an aggregate for which we created
                 SFTs, add the SFT corresponding to VI->OFFSET.  */
              tree sft = get_subvar_at (vi->decl, vi->offset);
              bitmap_set_bit (into, DECL_UID (sft));
            }
get_subvar_at (vi->decl, vi->offset) for
vi->decl global, vi->offset 96 is NULL.
static struct {
  int64_t s_f;
  rwlock_t s_f_lock;
  atomic_t s_f_lock_skip_count;
} global = {
  .s_f_lock_skip_count = { (0) },
};

create_overlap_variables_for only creates sft for s_f, because up_lookup says up->max_used == 64.

The http://gcc.gnu.org/ml/gcc-patches/2005-10/msg00057.html patch for 4.2 removed anyoffset handling (not desirable for 4.1 of course) and also without any comment added if (sft) right before the bitmap_set_bit, I wonder if that one liner couldn't be safe in this case.
Comment 2 Jakub Jelinek 2010-06-21 09:19:17 UTC 
Created attachment 425575 [details]
gcc41-rh605803.patch

Another, perhaps safer, version.  Instead of crashing just assume it can point to anything.
Comment 3 Jakub Jelinek 2010-06-21 09:20:47 UTC 
Created attachment 425577 [details]
gcc41-rh605803.patch

Another, perhaps safer, version.  Instead of crashing just assume it can point to anything.
Comment 5 Jakub Jelinek 2010-09-27 06:34:17 UTC 
In gcc-4.1.2-49.el5
Comment 10 errata-xmlrpc 2011-01-13 23:58:30 UTC 
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2011-0025.html