Re: [lmi] UBSAN flags

[Greg Chicares]

On 6/3/22 21:23, Vadim Zeitlin wrote:
> On Fri, 3 Jun 2022 19:51:19 +0000 Greg Chicares <gchicares@sbcglobal.net> 
> wrote:
> 
> GC> On 6/3/22 17:47, Vadim Zeitlin wrote:
> GC> > On Fri, 3 Jun 2022 16:29:18 +0000 Greg Chicares 
> <gchicares@sbcglobal.net> wrote:
> [...]
> GC> > GC> My first question is whether you ever got 'i7702.cpp' to finish
> GC> > GC> compiling. It's gone for many minutes here already, even with
> GC> > GC> the second (not '#'-commented) set of flags above.
> GC> > 
> GC> >  I didn't have any problem with this or any other file, but I didn't use
> GC> > -fno-sanitize-recover=all because by the time I learnt about it, I had
> GC> > already built everything without it and I didn't want to rebuild again.
> GC> 
> GC> What optimize option did you use? I used '-O3', and perhaps that's
> GC> my problem. I'll try again with '-O0' or perhaps '-Og'.
> 
>  I didn't use any optimization flags at all, i.e. the equivalent of "-O0",
> but "-Og" would be a better choice. I think UBSAN documentation recommends
> using at least "-O1" as the performance of the generated code is too poor
> otherwise, but at least for running the lmi unit tests "-O0" is good
> enough.

This attempt to compile only 'i7702.cpp' didn't look like
it would succeed, so I Ctrl-C'd it after 78 minutes.

Here's the command I used, but of course I've customized
the makefile...

/opt/lmi/src/lmi[0]$time make $coefficiency unit_tests 
unit_test_targets=i7702_test build_type=ubsan 2>&1 |less -S
make $coefficiency unit_tests unit_test_targets=i7702_test build_type=ubsan 2  
0.43s user 0.16s system 0% cpu 1:18:09.69 total
less -S  0.01s user 0.00s system 0% cpu 1:20:10.86 total

...so let me post the full command (everything that was
in the 'less' session started above):

g++ -MMD -MP -MT i7702.o -MF i7702.d  -c -I /opt/lmi/src/lmi -I 
/opt/lmi/src/lmi/tools/pete-2.1.1 -isystem 
/opt/lmi/local/gcc_x86_64-pc-linux-gnu/lib/wx/include/gtk3-unicode-3.1 -isystem 
/opt/lmi/local/include/wx-3.1 -isystem /opt/lmi/third_party/include -isystem 
/opt/lmi/local/include -isystem /opt/lmi/local/include/libxml2 
-DLMI_WX_NEW_USE_SO     -D_FILE_OFFSET_BITS=64 -DWXUSINGDLL -D__WXGTK__ 
-D_FILE_OFFSET_BITS=64   -fno-ms-extensions -frounding-math -std=c++20 
-pedantic-errors -Werror -Wall -Walloc-zero -Walloca -Wcast-align 
-Wcast-function-type -Wconversion -Wdangling-else -Wdeprecated-declarations 
-Wdisabled-optimization -Wdouble-promotion -Wduplicated-branches 
-Wduplicated-cond -Wextra -Wformat-nonliteral -Wformat-security 
-Wformat-signedness -Wformat-y2k -Wimport -Winvalid-pch -Wlogical-op 
-Wmissing-include-dirs -Wmultichar -Wnull-dereference -Wpacked -Wpointer-arith 
-Wredundant-decls -Wrestrict -Wshadow -Wsign-compare -Wstack-protector 
-Wswitch-enum -Wtrampolines -Wundef -Wunreachable-code -Wunused-macros 
-Wvector-operation-performance -Wno-parentheses  -Wc++11-compat -Wc++14-compat 
-Wc++1z-compat -Wcatch-value=3 -Wconditionally-supported -Wctor-dtor-privacy 
-Wdelete-non-virtual-dtor -Wdeprecated -Wextra-semi -Wnoexcept -Wnoexcept-type 
-Wnon-template-friend -Wnon-virtual-dtor -Wold-style-cast -Woverloaded-virtual 
-Wplacement-new=2 -Wpmf-conversions -Wregister -Wreorder -Wstrict-null-sentinel 
-Wsuggest-override -Wsynth -Wuseless-cast -Wzero-as-null-pointer-constant  
-Wcast-qual  -Wredundant-tags -Wvolatile     -O0 -fno-omit-frame-pointer -ggdb 
-fsanitize=undefined -fno-var-tracking-assignments -fPIC   
/opt/lmi/src/lmi/i7702.cpp -oi7702.o
make[2]: *** [/opt/lmi/src/lmi/workhorse.make:925: i7702.o] Interrupt
make[1]: [/opt/lmi/src/lmi/workhorse.make:1281: unit_tests] Interrupt (ignored)
make: *** [GNUmakefile:240: /opt/lmi/gcc_x86_64-pc-linux-gnu/build/ubsan] 
Interrupt

Other information:

/opt/lmi/src/lmi[0]$echo $LMI_TRIPLET 
x86_64-pc-linux-gnu

/opt/lmi/src/lmi[0]$gcc -dumpfullversion
11.2.0

The relevant parts seem to be:
  -O0: it worked for you
  -fno-omit-frame-pointer: so that UBSAN can give useful reports
  -fsanitize=undefined: sine qua non
  -fno-var-tracking-assignments: I hoped that would make it faster

I could also try
  -fno-sanitize-recover=all
but, as said above, UBSAN worked for you without it.

Am I missing something obvious?