Re: [PATCH] sim/m32c: fix memory leaks in opc2c

[Mike Frysinger via Gdb-patches <gdb-patches@sourceware.org>]

On 06 Apr 2021 21:45, Simon Marchi via Gdb-patches wrote:
> > i'm not keen on pushing it in this direction exactly.  it would mean every
> > caller would have to update & keep track.
> > 
> > i think you could define an IGEN variable in common/Make-common.in and change
> > all callers over to that.  then that would be the only place you'd have to add
> > any sanitizer related variables to.
> 
> I can do that, but I'm not sure where in common/Make-common.in I should
> add the variable.  There seems to be a logic to the organization in that
> file, but I don't get it.

*shrug* there isn't much to it.  you can put it after the POSTCOMPILE= line.

> Also, not that ppc has its own igen, so I
> guess it will still use its own definition.

correct, for now, it's duplicated and we eat that cost

> Another option is to just factor out the env var:
> 
>   DISABLE_LSAN = ASAN_OPTIONS=detect_leaks=0
> 
> ... and still use it in rules:
> 
> target: source
> 	$(DISABLE_LSAN) ../igen/igen --blah

i'm not keen on doing sanitizer-specific vars.  there's ASAN, LSAN, TSAN,
UBSAN, KSAN, and prob more in the future.  that's why having a common IGEN
var and then having it use a generic name (SANITIZE_ENV?) would work best
imo.

> Side-note, I saw some `@GMAKE_TRUE@` in Make-common.in.  In GDB, we
> decided to require GNU make and remove that complexity, in case you want
> to do the same.

sim/ is changing to automake, so a lot of that stuff will go away entirely.
i'm focusing on that rather than chipping away at smaller bits.

you can see this with the igen/ dir:
https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=b6b1c790843087e67e85e7cfd3327a872c03c6bc

although i need to do more groundwork in the C side first.
-mike