PBMS build failure in GCC 4.5
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Drizzle |
Fix Released
|
Medium
|
Barry Leslie | ||
7.0 |
Fix Released
|
Medium
|
Barry Leslie | ||
PBMS |
Fix Committed
|
Undecided
|
Barry Leslie |
Bug Description
With Fedora 14 due out early Nov I have been doing Drizzle testing on it. Looks like the clobber workaround isn't working in GCC 4.5:
cc1plus: warnings being treated as errors
plugin/
plugin/
make[2]: *** [plugin/
make[2]: *** Waiting for unfinished jobs....
make[2]: Leaving directory `/home/
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/
make: *** [all] Error 2
gcc --version
gcc (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
Related branches
- Mark Atwood: Needs Fixing
- Drizzle Merge Team: Pending requested
-
Diff: 43 lines (+1/-5)3 files modifiedplugin/pbms/src/backup_ms.cc (+0/-2)
plugin/pbms/src/cslib/CSStrUtil.h (+0/-1)
plugin/pbms/src/database_ms.cc (+1/-2)
- Lee Bieber (community): Needs Fixing
- Barry Leslie: Pending requested
-
Diff: 1378 lines (+175/-276)43 files modifieddrizzled/plugin/event_observer.cc (+3/-25)
drizzled/plugin/event_observer.h (+2/-2)
drizzled/session.cc (+6/-1)
plugin/hello_events/tests/r/bug616466_auto_increment_crash.result (+5/-0)
plugin/hello_events/tests/t/bug616466_auto_increment_crash.test (+5/-0)
plugin/pbms/src/backup_ms.cc (+6/-4)
plugin/pbms/src/connection_handler_ms.cc (+0/-1)
plugin/pbms/src/cslib/CSDefs.h (+1/-1)
plugin/pbms/src/cslib/CSEncode.cc (+1/-1)
plugin/pbms/src/cslib/CSMutex.cc (+19/-3)
plugin/pbms/src/cslib/CSPath.cc (+6/-2)
plugin/pbms/src/cslib/CSS3Protocol.cc (+9/-6)
plugin/pbms/src/cslib/CSStrUtil.cc (+1/-0)
plugin/pbms/src/cslib/CSStrUtil.h (+1/-0)
plugin/pbms/src/cslib/CSThread.cc (+3/-2)
plugin/pbms/src/database_ms.cc (+5/-6)
plugin/pbms/src/database_ms.h (+1/-1)
plugin/pbms/src/engine_ms.cc (+5/-4)
plugin/pbms/src/events_ms.cc (+1/-129)
plugin/pbms/src/ha_pbms.cc (+4/-3)
plugin/pbms/src/open_table_ms.cc (+1/-0)
plugin/pbms/src/open_table_ms.h (+1/-0)
plugin/pbms/src/pbms.h (+7/-6)
plugin/pbms/src/repository_ms.cc (+10/-10)
plugin/pbms/src/repository_ms.h (+3/-2)
plugin/pbms/src/systab_backup_ms.cc (+2/-2)
plugin/pbms/src/systab_backup_ms.h (+2/-2)
plugin/pbms/src/systab_cloud_ms.cc (+2/-2)
plugin/pbms/src/systab_cloud_ms.h (+2/-2)
plugin/pbms/src/systab_enabled_ms.cc (+2/-2)
plugin/pbms/src/systab_enabled_ms.h (+2/-2)
plugin/pbms/src/systab_httpheader_ms.cc (+2/-2)
plugin/pbms/src/systab_httpheader_ms.h (+2/-2)
plugin/pbms/src/systab_variable_ms.cc (+2/-2)
plugin/pbms/src/systab_variable_ms.h (+2/-2)
plugin/pbms/src/system_table_ms.cc (+9/-9)
plugin/pbms/src/system_table_ms.h (+8/-8)
plugin/pbms/src/trans_log_ms.cc (+14/-8)
plugin/pbms/src/transaction_ms.cc (+18/-0)
plugin/pbms/tests/r/basic.result (+0/-6)
plugin/pbms/tests/r/bug616466_auto_increment_crash.result (+0/-5)
plugin/pbms/tests/t/basic.test (+0/-5)
plugin/pbms/tests/t/bug616466_auto_increment_crash.test (+0/-6)
Changed in drizzle: | |
assignee: | nobody → Paul McCullagh (paul-mccullagh) |
Changed in drizzle: | |
assignee: | Paul McCullagh (paul-mccullagh) → Barry Leslie (barry-leslie) |
This is a bit of a bugger because I think this is actually a compiler/optimizer bug.
I can rearrange the code so that 'err' is not even in scope when setjmp() is called and it still reports it as possibly being clobbered.
I do not like the CLOBBER_PROTECT() macro though because it works by making assumptions about what the compiler does.
I will see if I cannot change this so that CLOBBER_PROTECT() is no longer needed.