Re: [PATCH v3 1/1] gdb/dwarf2: Fix 'rw_pieced_value' for values casted to different type.

[Simon Marchi <simark@simark.ca>]

On 1/5/23 14:52, Tom Tromey wrote:
>>>>>> Stephan Rohr via Gdb-patches <gdb-patches@sourceware.org> writes:
> 
>> From: "Rohr, Stephan" <stephan.rohr@intel.com>
>> The 'rw_pieced_value' function is executed when fetching a (lazy)
>> variable described by 'DW_OP_piece' or 'DW_OP_bit_piece'.  The
>> function checks the 'type' and 'enclosing_type' fields of the value
>> for identity.
> 
> Thanks for the patch.
> 
>> If a lazy value is fetched, GDB allocates space based on the enclosing
>> type's length and typically reads the 'full' object.  This is not
>> implemented for pieced values and causes an internal error if 'type'
>> and 'enclosing_type' of a value are not identical.
> 
>> However, GDB can read the value based on its type.  Thus, this patch
>> fixes the previously mentioned cases by removing the check for identity.
> 
> I thought there was some other discussion & idea about the cause of this
> patch... something like, setting a value's type should reset the
> enclosing type?  Or vice versa?  I am wondering if you tried this
> approach instead.

For reference, here was my last attempt at understanding what happens:

  https://inbox.sourceware.org/gdb-patches/dd04b3b4-ea6a-07ba-d734-0e542ca136b3@simark.ca/

I don't have a strong opinion on the matter, I don't know if one
solution is more corrent than the other.  The thing is, I can't explain
the logic behind the existing assert.  It's missing a comment explaining
why it would be a bad thing to "create a lazy value with an enclosing
type".  If we can't add such a justification, I think it's ok to remove
it...

Simon