[v2] docs: Suggest options to improve ASAN stack traces

On Thu, Dec 08, 2022 at 03:28:13PM +0100, Jakub Jelinek wrote:
> On Thu, Dec 08, 2022 at 09:11:54AM -0500, Marek Polacek via Gcc-patches wrote:
> > On Thu, Dec 08, 2022 at 08:25:26AM +0100, Florian Weimer wrote:
> > > * Marek Polacek via Gcc-patches:
> > > 
> > > > diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
> > > > index 726392409b6..2de14466dd3 100644
> > > > --- a/gcc/doc/invoke.texi
> > > > +++ b/gcc/doc/invoke.texi
> > > > @@ -16510,6 +16510,14 @@ The option cannot be combined with @option{-fsanitize=thread} or
> > > >  @option{-fsanitize=hwaddress}.  Note that the only target
> > > >  @option{-fsanitize=hwaddress} is currently supported on is AArch64.
> > > >  
> > > > +To get more accurate stack traces, it is possible to use options such as
> > > > +@option{-O} (which, for instance, prevents most function inlining),
> > > > +@option{-fno-optimize-sibling-calls} (which prevents optimizing sibling
> > > > +and tail recursive calls), or @option{-fno-ipa-icf} (which disables Identical
> > > > +Code Folding for functions and read-only variables).  Since multiple runs
> > > > +of the program may yield backtraces with different addresses due to ASLR,
> > > > +it may be desirable to turn off ASLR: @samp{setarch `uname -m` -R ./prog}.
> > > 
> > > What about -fasynchronous-unwind-tables?  It should help if ASAN ever
> > > reports stray segmentation faults.  Whether it also helps in general
> > > depends on whether ASAN maintains ABI around its instrumentation.
> > 
> > I'm not sure.  Someone else will have to decide if we want to mention
> > that option as well.
> 
> -fasynchronous-unwind-tables is on by default on many targets, so I wouldn't
> mention it:
> grep asynchronous_unwind_tables common/*/*/* config/*/*
> common/config/aarch64/aarch64-common.cc:    { OPT_LEVELS_ALL, OPT_fasynchronous_unwind_tables, NULL, 1 },
> common/config/i386/i386-common.cc:  opts->x_flag_asynchronous_unwind_tables = 2;
> common/config/loongarch/loongarch-common.cc:  { OPT_LEVELS_ALL, OPT_fasynchronous_unwind_tables, NULL, 1 },
> common/config/rs6000/rs6000-common.cc:  opts->x_flag_asynchronous_unwind_tables = 1;
> common/config/s390/s390-common.cc:  opts->x_flag_asynchronous_unwind_tables = 1;
> config/i386/i386-options.cc:      if (opts->x_flag_asynchronous_unwind_tables == 2)
> config/i386/i386-options.cc:	opts->x_flag_asynchronous_unwind_tables = !USE_IX86_FRAME_POINTER;
> config/mips/mips.cc:      && !global_options_set.x_flag_asynchronous_unwind_tables)
> config/mips/mips.cc:    flag_asynchronous_unwind_tables = 1;
> config/rs6000/rs6000.cc:      && !OPTION_SET_P (flag_asynchronous_unwind_tables))
> config/rs6000/rs6000.cc:    flag_asynchronous_unwind_tables = 1;
> 
> On the other side, the @samp{setarch `uname -m` -R ./prog} suggestion is
> very Linux specific, so if we mention it at all, it should mention that
> "e.g. on Linux through ..." or something similar.
> I also wouldn't mention the "and read-only variables" part, that is
> irrelevant for stack traces.

Thanks, updated patch here.  I've also expanded the ASLR acronym.

Ok?

-- >8 --
I got a complaint that while Clang docs suggest options that improve
the quality of the backtraces ASAN prints (cf.
<https://clang.llvm.org/docs/AddressSanitizer.html#usage>), our docs
don't say anything to that effect.  This patch amends that with a new
paragraph.  (It deliberately doesn't mention -fno-omit-frame-pointer.)

gcc/ChangeLog:

	* doc/invoke.texi (-fsanitize=address): Suggest options to improve
	stack traces.
---
 gcc/doc/invoke.texi | 9 +++++++++
 1 file changed, 9 insertions(+)


base-commit: d9f9d5d30feb33c359955d7030cc6be50ef6dc0a

On Thu, Dec 08, 2022 at 09:34:34AM -0500, Marek Polacek wrote:
> I got a complaint that while Clang docs suggest options that improve
> the quality of the backtraces ASAN prints (cf.
> <https://clang.llvm.org/docs/AddressSanitizer.html#usage>), our docs
> don't say anything to that effect.  This patch amends that with a new
> paragraph.  (It deliberately doesn't mention -fno-omit-frame-pointer.)
> 
> gcc/ChangeLog:
> 
> 	* doc/invoke.texi (-fsanitize=address): Suggest options to improve
> 	stack traces.
> ---
>  gcc/doc/invoke.texi | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
> index 726392409b6..1641efecf18 100644
> --- a/gcc/doc/invoke.texi
> +++ b/gcc/doc/invoke.texi
> @@ -16510,6 +16510,15 @@ The option cannot be combined with @option{-fsanitize=thread} or
>  @option{-fsanitize=hwaddress}.  Note that the only target
>  @option{-fsanitize=hwaddress} is currently supported on is AArch64.
>  
> +To get more accurate stack traces, it is possible to use options such as
> +@option{-O} (which, for instance, prevents most function inlining),

Still not sure about this part.  For one, I wonder if we shouldn't
recommend -O0, -O1 or -Og instead of just one of them, and I'm also not sure
how much function inlining is prevented with -O1.
always_inline functions are certainly inlined even at -O0 or -Og (at least
when called directly), -O1 adds
    { OPT_LEVELS_1_PLUS_NOT_DEBUG, OPT_finline_functions_called_once, NULL, 1 },
to that, -O2 adds
    { OPT_LEVELS_2_PLUS, OPT_findirect_inlining, NULL, 1 },
    { OPT_LEVELS_2_PLUS, OPT_finline_small_functions, NULL, 1 },
    { OPT_LEVELS_2_PLUS, OPT_fpartial_inlining, NULL, 1 },
    { OPT_LEVELS_2_PLUS, OPT_finline_functions, NULL, 1 },
and -O3 further bumps some parameters:
    { OPT_LEVELS_3_PLUS, OPT__param_max_inline_insns_auto_, NULL, 30 },
    { OPT_LEVELS_3_PLUS, OPT__param_early_inlining_insns_, NULL, 14 },
    { OPT_LEVELS_3_PLUS, OPT__param_inline_heuristics_hint_percent_, NULL, 600 },
    { OPT_LEVELS_3_PLUS, OPT__param_inline_min_speedup_, NULL, 15 },
    { OPT_LEVELS_3_PLUS, OPT__param_max_inline_insns_single_, NULL, 200 },

> +@option{-fno-optimize-sibling-calls} (which prevents optimizing sibling

-fno-optimize-sibling-calls is the default for -O0/-O1/-Og; dunno if we
want to reiterate it.

> +and tail recursive calls), or @option{-fno-ipa-icf} (which disables Identical
> +Code Folding for functions).  Since multiple runs of the program may yield
> +backtraces with different addresses due to ASLR (Address Space Layout
> +Randomization), it may be desirable to turn ASLR off.  On Linux, this can be
> +achieved with @samp{setarch `uname -m` -R ./prog}.
> +
>  @item -fsanitize=kernel-address
>  @opindex fsanitize=kernel-address
>  Enable AddressSanitizer for Linux kernel.

	Jakub