|
From: Roland M. <rol...@nr...> - 2013-08-09 16:36:17
|
Hi! ---- While testing ksh93 (ast-ksh.2013-08-07) under "valgrind" control (requires patch from http://lists.research.att.com/pipermail/ast-developers/2013q2/002574.html to deal with the custom (async-signal-safe) |malloc()| implementation) I hit the following warning: -- snip -- ==44984== Warning: ignored attempt to set SIGRT32 handler in sigaction(); ==44984== the SIGRT32 signal is used internally by Valgrind ==44984== Conditional jump or move depends on uninitialised value(s) ==44984== at 0x414910: sh_sigtrap (fault.c:318) ==44984== by 0x48BB46: b_trap (trap.c:170) ==44984== by 0x468E00: sh_exec (xec.c:1360) ==44984== by 0x46BEA6: sh_exec (xec.c:2223) ==44984== by 0x40F502: exfile (main.c:603) ==44984== by 0x40E74D: sh_main (main.c:375) ==44984== by 0x40D920: main (pmain.c:45) -- snip -- Erm... the issue here is that the POSIX realtime signals are "flexible", e.g. signal range they use is defined by libc functions. Is there any reason why valgrind can't intercept these functions and reduce the application-visible number of signals by one so it can use SIGRTMAX in peace ? ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) rol...@nr... \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 3992797 (;O/ \/ \O;) |