I have been playing around with the 1.3 port on my "broken" machine. Still no luck with sorting out the segfault. I have spent a fair amount of time in gdb and valgrind trying to track down the source of the trouble, but no luck. The best I can do when compiling with gcc is to use -O0, which produces an executable which does not segfault. I have tried gcc 4.4, 4.5, and 4.6, to no avail. I have wondered if it was something to do with my architecture; I am running an AMD E-350 fusion processor in my machine. In light of this, I tried using the appropriate -march setting with gcc 4.6, but again no luck. The only bright side of all of this was that I decided to compile with Clang, and it built nicely with only one warning (which I could not see the reason for): chain.c:189:50: warning: too many arguments in call to 'Chain_debug_msg' Chain_debug_msg(FuncStr, Op, Branch, Dir, Info); ~~~~~~~~~~~~~~~ ^ I will keep playing with this, to see if I can figure out the problem with gcc, but my hope is fading with regards to finding a solution. Rob
On Fri, Jun 10, 2011 at 11:42:53PM -0400, Rob S wrote:
I have been playing around with the 1.3 port on my "broken" machine. Still no luck with sorting out the segfault. I have spent a fair amount of time in gdb and valgrind trying to track down the source of the trouble, but no luck. The best I can do when compiling with gcc is to use -O0, which produces an executable which does not segfault. I have tried gcc 4.4, 4.5, and 4.6, to no avail. I have wondered if it was something to do with my architecture; I am running an AMD E-350 fusion processor in my machine. In light of this, I tried using the appropriate -march setting with gcc 4.6, but again no luck. The only bright side of all of this was that I decided to compile with Clang, and it built nicely with only one warning (which I could not see the reason for):
chain.c:189:50: warning: too many arguments in call to 'Chain_debug_msg' Chain_debug_msg(FuncStr, Op, Branch, Dir, Info); ~~~~~~~~~~~~~~~ ^
That's simple, just check the last repo! ;)
I will keep playing with this, to see if I can figure out the problem with gcc, but my hope is fading with regards to finding a solution.
Compile with debug info: CFLAGS="-g -O0" CXXFLAGS="-g -O0" ./configure --enable-ssl then run it under gdb: gdb ./dillo when it segfaults, look where it was: bt that's a good point to start. HTH -- Cheers Jorge.-
Rob wrote:
I have been playing around with the 1.3 port on my "broken" machine. Still no luck with sorting out the segfault. I have spent a fair amount of time in gdb and valgrind trying to track down the source of the trouble, but no luck. The best I can do when compiling with gcc is to use -O0, which produces an executable which does not segfault. I have tried gcc 4.4, 4.5, and 4.6, to no avail. I have wondered if it was something to do with my architecture; I am running an AMD E-350 fusion processor in my machine. In light of this, I tried using the appropriate -march setting with gcc 4.6, but again no luck.
Have you tried 1) adding MSG()s to watch interesting variables? 2) Turning on various warning options in gcc for issues that matter when optimization is used (overflow, for instance)?
participants (3)
-
corvid＠lavabit.com -
jcid＠dillo.org -
mr_semantics＠hotmail.com