Hi Simon,

yes, that's fine with me. Thanks for cleaning it up and pushing it.

For reference, I attached the adjusted patch with the compunit nullptr 
check removed, matching your suggestion.

Best,
Ronald

On 6/3/26 17:32, Simon Marchi wrote:
> On 6/3/26 8:11 AM, Ronald Hecht wrote:
>> |Hi, this is the patch from PR tdep/34198. It fixes an internal
>> segmentation fault in z80_skip_prologue when setting "break main" on
>> some Z80 binaries. "break *main" works, which points at the
>> prologue/SAL path. Regression-tested manually with the reproducer from
>> the PR. Thanks, Ronald
>> https://sourceware.org/bugzilla/show_bug.cgi?id=34198 |
>> diff --git a/gdb/z80-tdep.c b/gdb/z80-tdep.c
>> index f7e207d02ec3..f2e9b09e3b91 100644
>> --- a/gdb/z80-tdep.c
>> +++ b/gdb/z80-tdep.c
>> @@ -495,12 +495,17 @@ z80_skip_prologue (struct gdbarch *gdbarch, CORE_ADDR pc)
>>     if (prologue_end != 0)
>>       {
>>         struct symtab_and_line prologue_sal = find_sal_for_pc (func_addr, 0);
>> -      struct compunit_symtab *compunit = prologue_sal.symtab->compunit ();
>> -      const char *debug_format = compunit->debugformat ();
>> -
>> -      if (debug_format != NULL &&
>> -         !strncasecmp ("dwarf", debug_format, strlen("dwarf")))
>> -       return std::max (pc, prologue_end);
>> +      if (prologue_sal.symtab != nullptr)
>> +       {
>> +         struct compunit_symtab *compunit = prologue_sal.symtab->compunit ();
>> +         if (compunit != nullptr)
> symtab->compunit() can't return nullptr, so I wouldn't add this check
> (that method should be modified to return a reference, to make it
> clear).
>
> The patch looks good otherwise.  Would you be ok with me pushing your
> patch with that change?
>
> Simon