From: Andrew Burgess <aburgess@redhat.com>
To: Simon Marchi <simark@simark.ca>, gdb-patches@sourceware.org
Subject: Re: [PATCHv2] gdb/mi: fix use after free of frame_info causing spurious notifications
Date: Tue, 29 Mar 2022 11:10:18 +0100 [thread overview]
Message-ID: <87tubh9iid.fsf@redhat.com> (raw)
In-Reply-To: <052a55b2-7f6c-61f1-a4d8-a23a0a46126b@simark.ca>
Simon Marchi <simark@simark.ca> writes:
> On 2022-03-24 14:52, Andrew Burgess via Gdb-patches wrote:
>> Simon,
>>
>> Thanks for the feedback. I believe this addresses all your points.
>> What do you think?
>>
>> Thanks,
>> Andrew
>>
>> ---
>>
>> In commit:
>>
>> commit a2757c4ed693cef4ecc4dcdcb2518353eb6b3c3f
>> Date: Wed Mar 16 15:08:22 2022 +0000
>>
>> gdb/mi: consistently notify user when GDB/MI client uses -thread-select
>>
>> Changes were made to GDB to address some inconsistencies in when
>> notifications are sent from a MI terminal to a CLI terminal (when
>> multiple terminals are in use, see new-ui command).
>>
>> Unfortunately, in order to track when the currently selected frame has
>> changed, that commit grabs a frame_info pointer before and after an MI
>> command has executed, and compares the pointers to see if the frame
>> has changed.
>>
>> This is not safe.
>>
>> If the frame cache is deleted for any reason then the frame_info
>> pointer captured before the command started, is no longer valid, and
>> any comparisons based on that pointer are undefined.
>>
>> This was leading to random test failures for some folk, see:
>>
>> https://sourceware.org/pipermail/gdb-patches/2022-March/186867.html
>>
>> This commit changes GDB so we no longer hold frame_info pointers, but
>> instead store the frame_id and frame_level, this is safe even when the
>> frame cache is flushed.
>> ---
>> gdb/mi/mi-main.c | 43 +++++++++++++++++++++++++++++++++----------
>> 1 file changed, 33 insertions(+), 10 deletions(-)
>>
>> diff --git a/gdb/mi/mi-main.c b/gdb/mi/mi-main.c
>> index abd033b22ae..28d4afc28c0 100644
>> --- a/gdb/mi/mi-main.c
>> +++ b/gdb/mi/mi-main.c
>> @@ -1974,27 +1974,50 @@ struct user_selected_context
>> {
>> /* Constructor. */
>> user_selected_context ()
>> - : m_previous_ptid (inferior_ptid),
>> - m_previous_frame (deprecated_safe_get_selected_frame ())
>> - { /* Nothing. */ }
>> + : m_previous_ptid (inferior_ptid)
>> + {
>> + save_selected_frame (&m_previous_frame_id, &m_previous_frame_level);
>> + }
>>
>> /* Return true if the user selected context has changed since this object
>> was created. */
>> bool has_changed () const
>> {
>> - return ((m_previous_ptid != null_ptid
>> - && inferior_ptid != null_ptid
>> - && m_previous_ptid != inferior_ptid)
>> - || m_previous_frame != deprecated_safe_get_selected_frame ());
>> + /* Did the selected thread change? */
>> + if (m_previous_ptid != null_ptid && inferior_ptid != null_ptid
>> + && m_previous_ptid != inferior_ptid)
>> + return true;
>> +
>> + /* Grab details of the currently selected frame, for comparison. */
>> + frame_id current_frame_id;
>> + int current_frame_level;
>> + save_selected_frame (¤t_frame_id, ¤t_frame_level);
>> +
>> + /* Did the selected frame level change? */
>> + if (current_frame_level != m_previous_frame_level)
>> + return true;
>> +
>> + /* Did the selected frame id change? If the innermost frame is
>> + selected then the level will be -1, and the frame-id will be
>> + null_frame_id. As comparing null_frame_id with itself always
>> + reports not-equal, we only do the equality test if we have something
>> + other than the innermost frame selected. */
>> + if (current_frame_level != -1
>> + && !frame_id_eq (current_frame_id, m_previous_frame_id))
>> + return true;
>> +
>> + /* Nothing changed! */
>> + return false;
>> }
>> private:
>> /* The previously selected thread. This might be null_ptid if there was
>> no previously selected thread. */
>> ptid_t m_previous_ptid;
>>
>> - /* The previously selected frame. This might be nullptr if there was no
>> - previously selected frame. */
>> - frame_info *m_previous_frame;
>> + /* The previously selected frame. If the innermost frame is selected
>> + then the frame_id will be null_frame_id, and the level will be -1. */
>
> I noticed you changed this from "if no frame is currently selected" to
> "if the innermost frame is selected". I think that both can happen,
> actually:
>
> - if you execute an MI command while the target is running, there is no
> selected frame
> - if you execute an MI command while the innermost frame is selected
>
> So my understanding is that you could mention both (although it might
> become a bit redundant with the big comment on top of selected_frame_id,
> in frame.c).
>
> LGTM in any case.
I updated the comment as you suggested and pushed this patch to mater
and gdb-12-branch.
Thanks,
Andrew
next prev parent reply other threads:[~2022-03-29 10:10 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-16 15:09 [pushed] gdb/mi: consistently notify user when GDB/MI client uses -thread-select Jan Vrany
2022-03-21 20:35 ` Simon Marchi
2022-03-21 21:01 ` Jan Vrany
2022-03-22 14:37 ` Jan Vrany
2022-03-23 12:36 ` Simon Marchi
2022-03-23 15:13 ` Jan Vrany
2022-03-23 15:16 ` Simon Marchi
2022-03-23 16:09 ` Jan Vrany
2022-03-23 19:09 ` Andrew Burgess
2022-03-23 19:11 ` Andrew Burgess
2022-03-23 20:33 ` Jan Vrany
2022-03-24 14:47 ` Simon Marchi
2022-03-24 18:52 ` [PATCHv2] gdb/mi: fix use after free of frame_info causing spurious notifications Andrew Burgess
2022-03-24 20:32 ` Simon Marchi
2022-03-29 10:10 ` Andrew Burgess [this message]
2022-03-24 15:00 ` [pushed] gdb/mi: consistently notify user when GDB/MI client uses -thread-select Simon Marchi
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=87tubh9iid.fsf@redhat.com \
--to=aburgess@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=simark@simark.ca \
/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).