From: Simon Marchi <simon.marchi@polymtl.ca>
To: Bruno Larsen <blarsen@redhat.com>, gdb-patches@sourceware.org
Cc: Simon Marchi <simon.marchi@efficios.com>
Subject: Re: [PATCH 7/7] gdb: add special handling for frame level 0 in frame_info_ptr
Date: Tue, 8 Nov 2022 11:19:21 -0500 [thread overview]
Message-ID: <4ea9edd4-b268-5e8c-b697-b5dd09e2e25a@polymtl.ca> (raw)
In-Reply-To: <24a4edb5-4081-7b34-a5ca-291d51decea4@redhat.com>
On 11/8/22 05:40, Bruno Larsen wrote:
> On 07/11/2022 16:53, Simon Marchi via Gdb-patches wrote:
>> From: Simon Marchi <simon.marchi@efficios.com>
>>
>> I noticed this problem while preparing the initial submission for the
>> ROCm GDB port. One particularity of this patch set is that it does not
>> support unwinding frames, that requires support of some DWARF extensions
>> that will come later. It was still possible to run to a breakpoint and
>> print frame #0, though.
>>
>> When rebasing on top of the frame_info_ptr work, GDB started tripping on
>> a prepare_reinflate call, making it not possible anymore to event print
>> the frame when stopping on a breakpoint. One thing to know about frame
>> 0 is that its id is lazily computed when something requests it through
>> get_frame_id. See:
>>
>> https://gitlab.com/gnutools/binutils-gdb/-/blob/23912acd402f5af9caf91b257e5209ec4c58a09c/gdb/frame.c#L2070-2080
>>
>> So, up to that prepare_reinflate call, frame 0's id was not computed,
>> and prepare_reinflate, calling get_frame_id, forces it to be computed.
>> Computing the frame id generally requires unwinding the previous frame,
>> which with my ROCm GDB patch fails. An exception is thrown and the
>> printing of the frame is simply abandonned.
>>
>> Regardless of this ROCm GDB problem (which is admittedly temporary, it
>> will be possible to unwind with subsequent patches), we want to avoid
>> prepare_reinflate to force the computing of the frame id, for the same
>> reasons we lazily compute it in the first place.
>>
>> In addition, frame 0's id is subject to change across a frame cache
>> reset. This is why save_selected_frame and restore_selected_frame have
>> special handling for frame 0:
>>
>> https://gitlab.com/gnutools/binutils-gdb/-/blob/23912acd402f5af9caf91b257e5209ec4c58a09c/gdb/frame.c#L1841-1863
>>
>> For this last reason, we also need to handle frame 0 specially in
>> prepare_reinflate / reinflate. Because the frame id of frame 0 can
>> change across a frame cache reset, we must not rely on the frame id from
>> that frame to reinflate it. We should instead just re-fetch the current
>> frame at that point.
>>
>> This patch adds a frame_info_ptr::m_cached_level field, set in
>> frame_info_ptr::prepare_reinflate, so we can tell if a frame is frame 0.
>> There are cases where a frame_info_ptr object wraps a sentinel frame,
>> for which frame_relative_level returns -1, so I have chosen the value -2
>> to represent "invalid frame level", for when the frame_info_ptr object
>> is empty.
>>
>> In frame_info_ptr::prepare_reinflate, only cache the frame id if the
>> frame level is not 0. It's fine to cache the frame id for the sentinel
>> frame, it will be properly handled by frame_find_by_id later.
>>
>> In frame_info_ptr::reinflate, if the frame level is 0, call
>> get_current_frame to get the target's current frame. Otherwise, use
>> frame_find_by_id just as before.
>>
>> This patch should not have user-visible changes with upstream GDB. But
>> it will avoid forcing the computation of frame 0's when calling
>> prepare_reinflate. And, well, it fixes the upcoming ROCm GDB patch
>> series.
>>
>> Change-Id: I176ed7ee9317ddbb190acee8366e087e08e4d266
>
> This all makes sense. I have a small style preference below, but even if you dislike it, the code is still fine.
>
> Reviewed-By: Bruno Larsen <blarsen@redhat.com>
>
>> ---
>> gdb/frame-info.c | 24 ++++++++++++++++++++----
>> gdb/frame-info.h | 20 ++++++++++++++++++--
>> 2 files changed, 38 insertions(+), 6 deletions(-)
>>
>> diff --git a/gdb/frame-info.c b/gdb/frame-info.c
>> index 584222dc490f..e3ee9f8174e1 100644
>> --- a/gdb/frame-info.c
>> +++ b/gdb/frame-info.c
>> @@ -31,7 +31,11 @@ intrusive_list<frame_info_ptr> frame_info_ptr::frame_list;
>> void
>> frame_info_ptr::prepare_reinflate ()
>> {
>> - m_cached_id = get_frame_id (*this);
>> + m_cached_level = frame_relative_level (*this);
>> + gdb_assert (m_cached_level >= -1);
>
> Since you have declared invalid_level = -2 for this class, I feel like it would be more better to have the assert be
>
> gdb_assert (m_cached_level > invalid_level);
This form assumes that invalid_level is -2, defeating the purpose to
have the abstraction in the first place. If we changed invalid_level to
be INT_MAX, for intsance, the assertion wouldn't be right anymore.
> This way there is no need to wonder why -1 is a valid level, and makes it easier to grep for the comment in the file, should someone want to know.
In my vision of things, the sentinel frame having level -1 is well
known, because it's the frame just below the current frame, which is
known to have level 0. So while it looks like a magic random value,
it's not really. The numerical value has a meaning. We wouldn't want
to change the sentinel frame's level value to be any other arbitrary
numerical value.
Here, I can just drop the assert. It's basically just checking that
frame_relative_level didn't return something that doesn't make sense.
But there's no reason for frame_relative_level to return something that
doesn't make sense in the first place. Other callers of
frame_relative_level don't do this assert, they just trust that
frame_relative_level returns something that makes sense, nothing really
different here.
>> +
>> + if (m_cached_level != 0)
>> + m_cached_id = get_frame_id (*this);
>> }
>> /* See frame-info-ptr.h. */
>> @@ -39,9 +43,21 @@ frame_info_ptr::prepare_reinflate ()
>> void
>> frame_info_ptr::reinflate ()
>> {
>> - gdb_assert (frame_id_p (m_cached_id));
>> + gdb_assert (m_cached_level >= -1);
> Likewise
Here, I could add a comment like:
/* Ensure we have a valid frame level, indicating prepare_reinflate
was called. */
Simon
next prev parent reply other threads:[~2022-11-08 16:19 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-11-07 15:53 [PATCH 1/7] gdb: clear other.m_cached_id in frame_info_ptr's move ctor Simon Marchi
2022-11-07 15:53 ` [PATCH 2/7] gdb: add prepare_reinflate/reinflate around print_frame_args in info_frame_command_core Simon Marchi
2022-11-08 9:32 ` Bruno Larsen
2022-11-08 15:55 ` Simon Marchi
2022-11-08 19:39 ` Tom Tromey
2022-11-08 19:55 ` Simon Marchi
2022-11-07 15:53 ` [PATCH 3/7] gdb: move frame_info_ptr method implementations to frame-info.c Simon Marchi
2022-11-08 9:55 ` Bruno Larsen
2022-11-08 17:39 ` Tom Tromey
2022-11-07 15:53 ` [PATCH 4/7] gdb: remove manual frame_info reinflation code in backtrace_command_1 Simon Marchi
2022-11-08 10:14 ` Bruno Larsen
2022-11-08 16:05 ` Simon Marchi
2022-11-07 15:53 ` [PATCH 5/7] gdb: use frame_id_p instead of comparing to null_frame_id in frame_info_ptr::reinflate Simon Marchi
2022-11-08 17:43 ` Tom Tromey
2022-11-07 15:53 ` [PATCH 6/7] gdb: add missing prepare_reinflate call in print_frame_info Simon Marchi
2022-11-08 10:28 ` Bruno Larsen
2022-11-08 11:31 ` Lancelot SIX
2022-11-08 16:08 ` Simon Marchi
2022-11-07 15:53 ` [PATCH 7/7] gdb: add special handling for frame level 0 in frame_info_ptr Simon Marchi
2022-11-08 10:40 ` Bruno Larsen
2022-11-08 16:19 ` Simon Marchi [this message]
2022-11-10 16:28 ` Bruno Larsen
2022-11-10 16:30 ` Simon Marchi
2022-11-08 8:53 ` [PATCH 1/7] gdb: clear other.m_cached_id in frame_info_ptr's move ctor Bruno Larsen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4ea9edd4-b268-5e8c-b697-b5dd09e2e25a@polymtl.ca \
--to=simon.marchi@polymtl.ca \
--cc=blarsen@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=simon.marchi@efficios.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).