On Fri, Aug 02, 2024 at 07:38:05PM -0400, Aaron Merey wrote:
> From: Heather McIntyre <h...@rice.edu>
>
> * libdw/dwarf_formref_die.c (dwarf_formref_die): Add locking
> around call to __libdw_intern_next_unit.
> * libdw/libdwP.h (__libdw_dieabbrev): Add locking.
>
> Signed-off-by: Heather S. McIntyre <h...@rice.edu>
> Signed-off-by: Aaron Merey <ame...@redhat.com>
> Signed-off-by: Mark Wielaard <m...@klomp.org>
>
> v3 changes:
> Move dwarf_setalt.c changes to patch 5/9 v3.
> Set __libdw_dieabbrev result before releasing lock.
> ---
> libdw/dwarf_formref_die.c | 3 +++
> libdw/libdwP.h | 25 +++++++++++++++++++++----
> 2 files changed, 24 insertions(+), 4 deletions(-)
>
> diff --git a/libdw/dwarf_formref_die.c b/libdw/dwarf_formref_die.c
> index 48ba8194..69d1bf1c 100644
> --- a/libdw/dwarf_formref_die.c
> +++ b/libdw/dwarf_formref_die.c
> @@ -92,7 +92,10 @@ dwarf_formref_die (Dwarf_Attribute *attr, Dwarf_Die
> *result)
> bool scan_debug_types = false;
> do
> {
> + rwlock_wrlock(attr->cu->dbg->dwarf_lock);
> cu = __libdw_intern_next_unit (attr->cu->dbg, scan_debug_types);
> + rwlock_unlock(attr->cu->dbg->dwarf_lock);
> +
> if (cu == NULL)
> {
> if (scan_debug_types == false)
Aha, sorry, so there is extra locking around __libdw_intern_next_unit
here. Shouldn't this be combined with the next patch?
Thanks,
Mark
