[PATCH v8 1/5] gdb: make gdbarch store a vector of frame unwinders
Guinevere Larsen
guinevere@redhat.com
Tue Jan 14 20:34:55 GMT 2025
On 1/14/25 11:28 AM, Andrew Burgess wrote:
> Guinevere Larsen <guinevere@redhat.com> writes:
>
>> Before this commit, all frame unwinders would be stored in the obstack
>> of a gdbarch and accessed by using the registry system. This made for
>> unwieldy code, and unnecessarily complex logic in the frame_unwinder
>> implementation, along with making frame_unwind structs be unable to have
>> non-trivial destructors.
>>
>> Seeing as a future patch of this series wants to refactor the
>> frame_unwind struct to use inheritance, and we'd like to not restrict
>> the future derived classes on what destructors are allowed. In
>> preparation for that change, this commit changes the registry in gdbarch
>> to instead store an std::vector, which doesn't require using an obstack
>> and doesn't rely on a linked list.
>>
>> There should be no user-visible changes.
>>
>> Reviewed-by: Thiago Jung Bauermann <thiago.bauermann@linaro.org>
>> ---
>> gdb/frame-unwind.c | 107 +++++++++++++++------------------------------
>> 1 file changed, 36 insertions(+), 71 deletions(-)
>>
>> diff --git a/gdb/frame-unwind.c b/gdb/frame-unwind.c
>> index 352779fcdcc..e61f6244913 100644
>> --- a/gdb/frame-unwind.c
>> +++ b/gdb/frame-unwind.c
>> @@ -31,61 +31,42 @@
>> #include "cli/cli-cmds.h"
>> #include "inferior.h"
>>
>> -struct frame_unwind_table_entry
>> +/* Default sniffers, that must always be the first in the unwinder list,
>> + no matter the architecture. */
>> +static constexpr auto standard_unwinders =
> I'm not a huge fan of `auto` when the type is well known. My thinking
> is write once, read many times. So I'd rather have the type information
> available when I read the code. For me:
>
> static constexpr std::initializer_list<const frame_unwind *> standard_unwinders =
>
> tells me what's going on...
>
>> {
>> - const struct frame_unwind *unwinder;
>> - struct frame_unwind_table_entry *next;
>> + &dummy_frame_unwind,
>> + /* The DWARF tailcall sniffer must come before the inline sniffer.
>> + Otherwise, we can end up in a situation where a DWARF frame finds
>> + tailcall information, but then the inline sniffer claims a frame
>> + before the tailcall sniffer, resulting in confusion. This is
>> + safe to do always because the tailcall sniffer can only ever be
>> + activated if the newer frame was created using the DWARF
>> + unwinder, and it also found tailcall information. */
>> + &dwarf2_tailcall_frame_unwind,
>> + &inline_frame_unwind,
>> };
>>
>> -struct frame_unwind_table
>> -{
>> - struct frame_unwind_table_entry *list = nullptr;
>> - /* The head of the OSABI part of the search list. */
>> - struct frame_unwind_table_entry **osabi_head = nullptr;
>> -};
>> +/* If an unwinder should be prepended to the list, this is the
>> + index in which it should be inserted. */
>> +static constexpr int prepend_unwinder_index = standard_unwinders.size ();
>>
>> -static const registry<gdbarch>::key<struct frame_unwind_table>
>> +static const registry<gdbarch>::key<std::vector<const frame_unwind *>>
>> frame_unwind_data;
>>
>> -/* A helper function to add an unwinder to a list. LINK says where to
>> - install the new unwinder. The new link is returned. */
>> -
>> -static struct frame_unwind_table_entry **
>> -add_unwinder (struct obstack *obstack, const struct frame_unwind *unwinder,
>> - struct frame_unwind_table_entry **link)
>> -{
>> - *link = OBSTACK_ZALLOC (obstack, struct frame_unwind_table_entry);
>> - (*link)->unwinder = unwinder;
>> - return &(*link)->next;
>> -}
>> -
>> -static struct frame_unwind_table *
>> +/* Retrieve the list of frame unwinders available in GDBARCH.
>> + If this list is empty, it is initialized before being returned. */
>> +static std::vector<const frame_unwind *> *
>> get_frame_unwind_table (struct gdbarch *gdbarch)
> Given you're changing this code anyway, how about returning a reference
> rather than a pointer, i.e.:
>
> static std::vector<const frame_unwind *> &
> get_frame_unwind_table (struct gdbarch *gdbarch)
> { ... }
>
> The users of get_frame_unwind_table will need to be updated to match.
>
>> {
>> - struct frame_unwind_table *table = frame_unwind_data.get (gdbarch);
>> + std::vector<const frame_unwind *> *table = frame_unwind_data.get (gdbarch);
>> if (table != nullptr)
>> return table;
>>
>> - table = new frame_unwind_table;
>> -
>> - /* Start the table out with a few default sniffers. OSABI code
>> - can't override this. */
>> - struct frame_unwind_table_entry **link = &table->list;
>> + table = new std::vector<const frame_unwind *>;
>> + table->insert (table->begin (), standard_unwinders.begin (),
>> + standard_unwinders.end ());
>>
>> - struct obstack *obstack = gdbarch_obstack (gdbarch);
>> - link = add_unwinder (obstack, &dummy_frame_unwind, link);
>> - /* The DWARF tailcall sniffer must come before the inline sniffer.
>> - Otherwise, we can end up in a situation where a DWARF frame finds
>> - tailcall information, but then the inline sniffer claims a frame
>> - before the tailcall sniffer, resulting in confusion. This is
>> - safe to do always because the tailcall sniffer can only ever be
>> - activated if the newer frame was created using the DWARF
>> - unwinder, and it also found tailcall information. */
>> - link = add_unwinder (obstack, &dwarf2_tailcall_frame_unwind, link);
>> - link = add_unwinder (obstack, &inline_frame_unwind, link);
>> -
>> - /* The insertion point for OSABI sniffers. */
>> - table->osabi_head = link;
>> frame_unwind_data.set (gdbarch, table);
>>
>> return table;
>> @@ -95,27 +76,16 @@ void
>> frame_unwind_prepend_unwinder (struct gdbarch *gdbarch,
>> const struct frame_unwind *unwinder)
>> {
>> - struct frame_unwind_table *table = get_frame_unwind_table (gdbarch);
>> - struct frame_unwind_table_entry *entry;
>> -
>> - /* Insert the new entry at the start of the list. */
>> - entry = GDBARCH_OBSTACK_ZALLOC (gdbarch, struct frame_unwind_table_entry);
>> - entry->unwinder = unwinder;
>> - entry->next = (*table->osabi_head);
>> - (*table->osabi_head) = entry;
>> + std::vector<const frame_unwind *> *table = get_frame_unwind_table (gdbarch);
>> +
>> + table->insert (table->begin () + prepend_unwinder_index, unwinder);
>> }
>>
>> void
>> frame_unwind_append_unwinder (struct gdbarch *gdbarch,
>> const struct frame_unwind *unwinder)
>> {
>> - struct frame_unwind_table *table = get_frame_unwind_table (gdbarch);
>> - struct frame_unwind_table_entry **ip;
>> -
>> - /* Find the end of the list and insert the new entry there. */
>> - for (ip = table->osabi_head; (*ip) != NULL; ip = &(*ip)->next);
>> - (*ip) = GDBARCH_OBSTACK_ZALLOC (gdbarch, struct frame_unwind_table_entry);
>> - (*ip)->unwinder = unwinder;
>> + get_frame_unwind_table (gdbarch)->push_back (unwinder);
>> }
>>
>> /* Call SNIFFER from UNWINDER. If it succeeded set UNWINDER for
>> @@ -188,9 +158,6 @@ frame_unwind_find_by_frame (const frame_info_ptr &this_frame, void **this_cache)
>> FRAME_SCOPED_DEBUG_ENTER_EXIT;
>> frame_debug_printf ("this_frame=%d", frame_relative_level (this_frame));
>>
>> - struct gdbarch *gdbarch = get_frame_arch (this_frame);
>> - struct frame_unwind_table *table = get_frame_unwind_table (gdbarch);
>> - struct frame_unwind_table_entry *entry;
>> const struct frame_unwind *unwinder_from_target;
>>
>> unwinder_from_target = target_get_unwinder ();
>> @@ -205,8 +172,10 @@ frame_unwind_find_by_frame (const frame_info_ptr &this_frame, void **this_cache)
>> unwinder_from_target))
>> return;
>>
>> - for (entry = table->list; entry != NULL; entry = entry->next)
>> - if (frame_unwind_try_unwinder (this_frame, this_cache, entry->unwinder))
>> + struct gdbarch *gdbarch = get_frame_arch (this_frame);
>> + std::vector<const frame_unwind *> *table = get_frame_unwind_table (gdbarch);
>> + for (auto unwinder : *table)
> I think you can use:
>
> for (const auto &unwinder : *table)
>
> here. The 'const' is just good style as you're not going to modify it.
> The '&' is not really necessary as the type is actually a pointer.
> Better still might be:
>
> for (const frame_unwind *unwinder : *table)
>
> But I don't mind the 'auto' here too much as the type is on the line
> immediately above. But I do think, if you're sticking with 'auto' here
> then throwing the '&' in is a good idea.
I decided to go with the "auto" version, since table is always declared
close to the usage, and that doesn't use auto.
>
>> + if (frame_unwind_try_unwinder (this_frame, this_cache, unwinder))
>> return;
>>
>> internal_error (_("frame_unwind_find_by_frame failed"));
>> @@ -347,7 +316,7 @@ static void
>> maintenance_info_frame_unwinders (const char *args, int from_tty)
>> {
>> gdbarch *gdbarch = current_inferior ()->arch ();
>> - struct frame_unwind_table *table = get_frame_unwind_table (gdbarch);
>> + std::vector<const frame_unwind *> *table = get_frame_unwind_table (gdbarch);
>>
>> ui_out *uiout = current_uiout;
>> ui_out_emit_table table_emitter (uiout, 2, -1, "FrameUnwinders");
>> @@ -355,15 +324,11 @@ maintenance_info_frame_unwinders (const char *args, int from_tty)
>> uiout->table_header (25, ui_left, "type", "Type");
>> uiout->table_body ();
>>
>> - for (struct frame_unwind_table_entry *entry = table->list; entry != NULL;
>> - entry = entry->next)
>> + for (auto unwinder : *table)
> See the comments above.
>
> If you're happy making this changes I suggest then:
>
> Approved-By: Andrew Burgess <aburgess@redhat.com>
Thanks for the review, I went along with the changes and added your
approval!
--
Cheers,
Guinevere Larsen
She/Her/Hers
>
> Thanks,
> Andrew
>
>> {
>> - const char *name = entry->unwinder->name;
>> - const char *type = frame_type_str (entry->unwinder->type);
>> -
>> ui_out_emit_list tuple_emitter (uiout, nullptr);
>> - uiout->field_string ("name", name);
>> - uiout->field_string ("type", type);
>> + uiout->field_string ("name", unwinder->name);
>> + uiout->field_string ("type", frame_type_str (unwinder->type));
>> uiout->text ("\n");
>> }
>> }
>> --
>> 2.47.0
More information about the Gdb-patches
mailing list