testing/gzip
SHA256
3
0
Fork 0
forked from pool/gzip
Code Pull Requests Activity

Accepting request 901465 from home:fbui:branches:Base:System

Browse Source 
- Avoid text relocation on i386 (bsc#1143125)
  Text relocation doesn't cope well with some security hardening which
  consists in prohibiting attempts to change memory mappings to become
  executable.
  Hence disable the use of assembler code and get rid of text
  relocations especially since the assembler code is slower than the
  code generated by gcc according to Fedora. According to Debian this
  also fixes gzip generating different results on ia64.

OBS-URL: https://build.opensuse.org/request/show/901465
OBS-URL: https://build.opensuse.org/package/show/Base:System/gzip?expand=0&rev=82
This commit is contained in:
Marcus Meissner 2021-06-29 11:57:07 +00:00 committed by Git OBS Bridge
parent 4dae0a7744
commit 3b3197d579
2 changed files with 17 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
gzip.changes
View File

@ -1,3 +1,17 @@
-------------------------------------------------------------------
Wed Jun 23 09:25:09 UTC 2021 - Franck Bui <fbui@suse.com>
- Avoid text relocation on i386 (bsc#1143125)
Text relocation doesn't cope well with some security hardening which
consists in prohibiting attempts to change memory mappings to become
executable.
Hence disable the use of assembler code and get rid of text
relocations especially since the assembler code is slower than the
code generated by gcc according to Fedora. According to Debian this
also fixes gzip generating different results on ia64.
-------------------------------------------------------------------
Tue Apr 13 13:23:44 UTC 2021 - pgajdos@suse.com

3
gzip.spec
View File

@ -76,6 +76,9 @@ export LDFLAGS="-pie"
autoreconf -f -i
export CFLAGS="$CFLAGS -DDFLTCC_LEVEL_MASK=0x7e"
%endif
# Avoid text relocations on i386 as the assembler code (in
# lib/match.c) is not prepared for PIE (bsc#1143125).
export DEFS=NO_ASM
%configure --disable-silent-rules \
gl_cv_func_printf_directive_n=yes \
gl_cv_func_printf_infinite_long_double=yes \