Re: [PATCH v5 0/2] gdb: Fix internal inferior calls through GNU IFUNCs

On Tue, 11 Aug 2026 13:12:17 +0000
Muhammad Kamran <[email protected]> wrote:

> This series fixes a GDB inferior-call issue exposed by malloc being a GNU
> IFUNC in glibc on AArch64.  The underlying problem is not AArch64-specific:
> it can affect any inferior helper found through find_function_in_inferior's
> minimal-symbol fallback when that helper is a GNU IFUNC.
> 
> GDB calls find_function_in_inferior ("malloc") when expression evaluation
> needs to allocate memory in the inferior, for example for string literal
> arguments.  In the minimal-symbol fallback, GDB created a synthetic ordinary
> function pointer from the minimal symbol address.  If the symbol was a GNU
> IFUNC, this lost the IFUNC marker, so call_function_by_hand did not resolve
> the symbol before calling it.
> 
> Patch 1 checks the minimal symbol kind directly and propagates the GNU IFUNC
> marker to the synthetic function type for mst_text_gnu_ifunc and
> mst_data_gnu_ifunc symbols.  The existing fallback address and return type
> are unchanged.
> 
> Patch 2 fixes the follow-on return-type issue after IFUNC resolution.  When
> the resolved target type, or the type inferred from the resolver return type,
> does not provide a usable return type, find_function_addr now keeps the
> original function type's return type.  This preserves the synthetic fallback
> return type used by find_function_in_inferior ("malloc").
> 
> The tests add an internal inferior-call case using an IFUNC malloc and run it
> through the existing gdb.base/gnu-ifunc.exp matrix for resolver attr,
> resolver debug info, and resolved-target debug info.
> 
> Changes since v4:
> 	* Address feedback from Andrew Burgess <[email protected]>.
> 
> Changes since v3:
> 	* Move the ifunc resolver into a shared library.
> 
> Changes since v2:
> 	* Rework the minimal-symbol fallback to check the minimal symbol kind
> 	  directly as suggested by Simon.
> 	* Add coverage for both debug and no-debug IFUNC malloc variants by
> 	  using the existing framework.
> 	* Preserve the original IFUNC return type when the resolved target type
> 	  is unknown.

Andrew gave conditional approval for both parts in his v4 replies. 
I've looked over this v5 series; IMO, you've completely addressed his
concerns.  Therefore, I think you can consider this version approved;
but do add Andrew's Approved-By tag.  If you need a maintainer to push
it for you, I'll be happy to do that.

Reviewed-By: Kevin Buettner <[email protected]>