From: Luis Machado <luis.machado@arm.com>
To: Pedro Alves <pedro@palves.net>,
Thiago Jung Bauermann <thiago.bauermann@linaro.org>
Cc: Simon Marchi <simark@simark.ca>,
Andrew Burgess <aburgess@redhat.com>,
Thiago Jung Bauermann via Gdb-patches
<gdb-patches@sourceware.org>
Subject: Re: [PATCH v3 4/8] gdbserver/linux-aarch64: When thread stops, update its target description
Date: Tue, 7 Feb 2023 08:11:19 +0000 [thread overview]
Message-ID: <d3c89dc9-5c2a-be4b-b0ed-0ce19ce8a007@arm.com> (raw)
In-Reply-To: <045c1e09-5b6d-22b6-df7a-39cfa339b0e1@palves.net>
On 2/6/23 20:29, Pedro Alves wrote:
> On 2023-02-04 3:21 p.m., Thiago Jung Bauermann via Gdb-patches wrote:
>>
>> Luis Machado <luis.machado@arm.com> writes:
>>
>>> On 2/2/23 03:47, Simon Marchi wrote:
>>>> On 2/1/23 21:54, Thiago Jung Bauermann wrote:
>>>>> In any case, it wouldn't be possible to make get_thread_target_desc just
>>>>> return thread_info->tdesc because at least the way these patches are
>>>>> currently written, when the inferior starts or a new thread of the
>>>>> inferior is spawned thread_info->tdesc is nullptr. gdbserver will only
>>>>> call get_thread_tdesc after the first stop (in get_thread_regcache, in
>>>>> the process of obtaining the pc register), so we will need to cope with
>>>>> that situation.
>>>> Ok. Would it work if a new thread initially inherited the tdesc from
>>>> its process?
>>>>
>>>
>>> It should be fine because the first time we fetch a process target
>>> description, it is eventually obtained from the first and only thread.
>>> So the SVE vector length should be correct.
>>>
>>> Any subsequent attempts to use the process' target description (the
>>> first one we obtained), after further stops, may end up using an
>>> incorrect description.
>>>
>>> I think this is handled correctly by the target architecture target
>>> hook though. But there are other places where this is potentially
>>> incorrect.
>>>
>>> For example...
>>>
>>> - When using gcore to dump a core file, GDB only dumps a single target
>>> description. While this might be correct for a target with a fixed
>>> target description or a AArch64 target that doesn't support SVE, it
>>> likely won't be correctly for one AArch64 target supporting SVE if its
>>> threads changed vector length mid-execution. Either we emit target
>>> description notes by thread, or we don't emit a target description
>>> note for those cases.
>>>
>>> - When loading the above/older gcore core files back, GDB will use a
>>> potentially incorrect target description. If we decide to emit
>>> per-thread target descriptions, it should be fine. Otherwise we may
>>> need to have a "thread architecture" hook for core files as well.
>>>
>>> - The remote has no concept of a thread architecture (Thiago is
>>> addressing this with this patch series).
>>>
>>> - AArch64 frames may have slightly different vg values, which means
>>> their gdbarches are different as well.
>>>
>>> Given the differences between two gdbarches are small, we mostly get
>>> away with it. But if there are further differences (different hooks,
>>> for example), I fear we may run into a situation where we use an
>>> incorrect gdbarch to call a particular hook.
>>
>> Indeed, good points! Thank you for bringing them up. I can address core
>> file dumping/loading after this series.
>>
>> Regarding frames with different vg values, it's important to be aware of
>> this discrepancy but IMHO it makes sense to work on it when it becomes
>> a problem...
>>
>
>
> Yeah, one thought that keeps crossing my mind is if really modeling all this
> stuff as different target descriptions is really the best approach. The Intel AMX
> support posted on the list last year also ran into a similar problem, with the
> matrix registers height/width changing at runtime, and it is impractical (or really,
> it really smells like the wrong approach) to have different target descriptions for
> every potential matrix size. Which is not unlike different SVE sizes. It feels like
> a single tdesc should be able to be a bit more dynamic. It's a bit funny that ptrace
> manages to work with a single registers interface, while we don't.
My understanding is that this was a bit hard to justify back when SVE was implemented, as it would
have to touch some other bits of the type system to create a sizeless type for SVE vectors, where
the size would be determined by context.
>
> What if we extended the target description mechanism so that a single description
> could describe all SVE sizes? For example, what if a tdesc could describe the SVE
> width/length as a dynamic property, retrieved from elsewhere, e.g., from another register?
>
> BTW, for core files, where are we going to retrieve the SVE length from?
The thread-specific vector length should be available from the SVE register dump sections.
next prev parent reply other threads:[~2023-02-07 8:11 UTC|newest]
Thread overview: 94+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-30 4:45 [PATCH v3 0/8] gdbserver improvements for AArch64 SVE support Thiago Jung Bauermann
2023-01-30 4:45 ` [PATCH v3 1/8] gdbserver: Add assert in find_register_by_number Thiago Jung Bauermann
2023-01-31 17:05 ` Simon Marchi
2023-01-31 19:49 ` Thiago Jung Bauermann
2023-02-01 15:43 ` Simon Marchi
2023-01-30 4:45 ` [PATCH v3 2/8] gdbserver: Add PID parameter to linux_get_auxv and linux_get_hwcap Thiago Jung Bauermann
2023-02-01 9:07 ` Luis Machado
2023-02-01 10:54 ` Andrew Burgess
2023-02-01 16:01 ` Simon Marchi
2023-02-01 19:33 ` Thiago Jung Bauermann
2023-02-01 19:53 ` Simon Marchi
2023-02-01 21:55 ` Thiago Jung Bauermann
2023-02-06 19:54 ` Pedro Alves
2023-02-06 20:16 ` Simon Marchi
2023-02-07 15:19 ` Pedro Alves
2023-02-07 21:47 ` Thiago Jung Bauermann
2023-02-09 1:31 ` Simon Marchi
2023-02-10 3:54 ` Thiago Jung Bauermann
2023-02-07 22:28 ` Thiago Jung Bauermann
2023-01-30 4:45 ` [PATCH v3 3/8] gdbserver/linux-aarch64: Factor out function to get aarch64_features Thiago Jung Bauermann
2023-02-01 8:59 ` Luis Machado
2023-02-01 16:04 ` Simon Marchi
2023-02-01 22:13 ` Thiago Jung Bauermann
2023-01-30 4:45 ` [PATCH v3 4/8] gdbserver/linux-aarch64: When thread stops, update its target description Thiago Jung Bauermann
2023-02-01 9:05 ` Luis Machado
2023-02-01 11:06 ` Andrew Burgess
2023-02-01 16:21 ` Simon Marchi
2023-02-01 16:32 ` Luis Machado
2023-02-02 2:54 ` Thiago Jung Bauermann
2023-02-02 3:47 ` Simon Marchi
2023-02-03 3:47 ` Thiago Jung Bauermann
2023-02-03 11:13 ` Luis Machado
2023-02-04 15:26 ` Thiago Jung Bauermann
2023-02-03 11:11 ` Luis Machado
2023-02-04 15:21 ` Thiago Jung Bauermann
2023-02-06 9:07 ` Luis Machado
2023-02-06 12:15 ` Thiago Jung Bauermann
2023-02-06 20:29 ` Pedro Alves
2023-02-07 8:11 ` Luis Machado [this message]
2023-02-07 14:39 ` Thiago Jung Bauermann
2023-02-03 10:57 ` Luis Machado
2023-02-04 6:18 ` Thiago Jung Bauermann
2023-02-06 20:26 ` Pedro Alves
2023-02-07 21:06 ` Thiago Jung Bauermann
2023-02-09 2:46 ` Simon Marchi
2023-02-10 3:29 ` Thiago Jung Bauermann
2023-02-10 14:56 ` Luis Machado
2023-02-10 15:04 ` Tom Tromey
2023-02-10 15:28 ` Luis Machado
2023-02-10 17:26 ` Thiago Jung Bauermann
2023-02-10 21:01 ` Simon Marchi
2023-01-30 4:45 ` [PATCH v3 5/8] gdbserver: Transmit target description ID in thread list and stop reply Thiago Jung Bauermann
2023-01-30 12:52 ` Eli Zaretskii
2023-01-30 14:05 ` Thiago Jung Bauermann
2023-02-01 9:39 ` Luis Machado
2023-02-01 12:07 ` Andrew Burgess
2023-02-01 13:03 ` Eli Zaretskii
2023-02-01 17:37 ` Simon Marchi
2023-02-02 20:36 ` Thiago Jung Bauermann
2023-02-02 20:56 ` Simon Marchi
2023-02-01 20:46 ` Simon Marchi
2023-02-02 21:43 ` Thiago Jung Bauermann
2023-02-01 14:51 ` Andrew Burgess
2023-02-01 17:03 ` Simon Marchi
2023-02-02 19:52 ` Thiago Jung Bauermann
2023-02-02 20:51 ` Simon Marchi
2023-02-03 2:44 ` Thiago Jung Bauermann
2023-02-03 16:29 ` Andrew Burgess
2023-02-04 6:08 ` Thiago Jung Bauermann
2023-02-03 11:22 ` Luis Machado
2023-02-03 12:50 ` Simon Marchi
2023-01-30 4:45 ` [PATCH v3 6/8] gdb/remote: Parse tdesc field in stop reply and threads list XML Thiago Jung Bauermann
2023-02-01 9:52 ` Luis Machado
2023-02-05 0:06 ` Thiago Jung Bauermann
2023-02-06 9:10 ` Luis Machado
2023-02-01 14:32 ` Andrew Burgess
2023-02-01 19:50 ` Simon Marchi
2023-02-01 20:16 ` Simon Marchi
2023-02-03 11:27 ` Luis Machado
2023-02-03 13:19 ` Simon Marchi
2023-02-03 16:33 ` Andrew Burgess
2023-01-30 4:45 ` [PATCH v3 7/8] gdb/aarch64: Detect vector length changes when debugging remotely Thiago Jung Bauermann
2023-02-01 9:58 ` Luis Machado
2023-02-01 15:26 ` Andrew Burgess
2023-02-01 20:20 ` Simon Marchi
2023-02-03 11:31 ` Luis Machado
2023-02-03 16:38 ` Andrew Burgess
2023-02-03 19:07 ` Simon Marchi
2023-01-30 4:45 ` [PATCH v3 8/8] gdb/testsuite: Add test to exercise multi-threaded AArch64 SVE inferiors Thiago Jung Bauermann
2023-02-01 10:10 ` Luis Machado
2023-02-06 19:11 ` [PATCH v3 0/8] gdbserver improvements for AArch64 SVE support Pedro Alves
2023-02-06 20:05 ` Simon Marchi
2023-02-06 21:06 ` Pedro Alves
2023-02-07 13:49 ` 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=d3c89dc9-5c2a-be4b-b0ed-0ce19ce8a007@arm.com \
--to=luis.machado@arm.com \
--cc=aburgess@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=pedro@palves.net \
--cc=simark@simark.ca \
--cc=thiago.bauermann@linaro.org \
/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).