Re: [PATCH] [gdb/exp] Fix printing of out of bounds struct members

[Tom de Vries <tdevries@suse.de>]

On 1/22/24 17:29, Tom Tromey wrote:
>>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:
> 
> Tom> When building gdb with -O0 -fsanitize=address, and running test-case
> Tom> gdb.ada/uninitialized_vars.exp, I run into:
> Tom> ...
> 
> Thanks for looking into this.
> 
> Tom> - when y.a happens to be set to 216, as above, gdb interprets this as the
> Tom>   variable having the c+d variant (which is why trying to print y.b fails).
> 
> One thing I see here is that y.a is not one of the valid values.  So I
> wonder if this kind of corruption should be detected at a different
> spot, like throwing an exception from compute_variant_fields_inner.
> 

I've looked into that, but I think it requires a cleanup related to 
TYPE_CODE_BOOL, see this PR ( 
https://sourceware.org/bugzilla/show_bug.cgi?id=31282 ).  Currently 
there's no reliable way to determine what valid bool values are.  I've 
updated the commit message to mention this way of fixing it in the v2.

> Tom> [ Given the description at the top of gdb/value.h of optimized out and
> Tom> unavailable, I was not sure which one to pick for this case, so this is more
> Tom> or less a random choice.
> 
> Yeah, neither seems like precisely the right fit.

OK, I've updated v2 to use an error then.

V2: https://sourceware.org/pipermail/gdb-patches/2024-January/206200.html

Thanks,
- Tom