This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH v2 1/2] compile: set debug compile: Display GCC driver filename
- From: Jan Kratochvil <jan dot kratochvil at redhat dot com>
- To: Pedro Alves <palves at redhat dot com>
- Cc: gdb-patches at sourceware dot org, Phil Muldoon <pmuldoon at redhat dot com>
- Date: Mon, 27 Apr 2015 22:26:03 +0200
- Subject: Re: [PATCH v2 1/2] compile: set debug compile: Display GCC driver filename
- Authentication-results: sourceware.org; auth=none
- References: <20150423203402 dot 23140 dot 92757 dot stgit at host1 dot jankratochvil dot net> <553E5646 dot 8020708 at redhat dot com> <20150427164757 dot GA10548 at host1 dot jankratochvil dot net> <553E6F92 dot 4020204 at redhat dot com> <20150427175213 dot GA12596 at host1 dot jankratochvil dot net> <553E85BA dot 8020107 at redhat dot com>
On Mon, 27 Apr 2015 20:53:46 +0200, Pedro Alves wrote:
> This split is a conscious design decision,
> that has advantages like insulating the debugger from compiler ICEs.
It would be useful if it was optional. This way it complicates the debugging
a lot. AFAIK it is not optional due to packaging reasons, cc1 would need to
be -fPIC which has some performance hit. OTOH that performance hit on x86_64
is say 1% (just my rough guess) but the performance hit of using GCC instead
of Clang is 119%. So I do not see why the -fPIC is not acceptable for GCC.
> And please don't tell me that LLDB/Clang don't crash. ;-)
I find it offtopic here but when you ask all software has bugs, for LLDB/clang:
* clang has not yet crashed for me on Linux during its daily use.
(gcc did but this is just due to using gcc longer / more extensively.)
* lldb has not crashed for me on OSX but I did not test it much there.
* lldb Linux port did crash for me but I do not consider that port finished.
Jan