On Sat, Jun 24, 2023, at 10:50, Michael Büsch wrote:
> On Sat, 24 Jun 2023 08:44:15 +0700
> Bagas Sanjaya <[email protected]> wrote:
>> > I suspect change introduced when addressing a compiler warning
>> > cased the error.
>> >
>> > https://patchwork.kernel.org/project/linux-wireless/patch/20230516183442.536589-1-arnd%40kernel.org/
>
>
> I doubt it.
> This patch affects the device initialization code. But the crash is in
> the transmit path.
> Can you please double check by manually reverting the patch?
I'm travelling at the moment and can't easily check it, but I would
expect that my patch has no effect on the generated object code at
all. If the compiler output is different with and without my patch,
it's probably wrong. I double-checked the structure layout in
https://godbolt.org, this did not produce any changes, as size,
alignement and offsets of the members are all the same.
Arnd
On 24.06.23 11:29, Arnd Bergmann wrote:
> On Sat, Jun 24, 2023, at 10:50, Michael Büsch wrote:
>> On Sat, 24 Jun 2023 08:44:15 +0700
>> Bagas Sanjaya <[email protected]> wrote:
>>>> I suspect change introduced when addressing a compiler warning
>>>> cased the error.
>>>>
>>>> https://patchwork.kernel.org/project/linux-wireless/patch/20230516183442.536589-1-arnd%40kernel.org/
>>
>> I doubt it.
>> This patch affects the device initialization code. But the crash is in
>> the transmit path.
>> Can you please double check by manually reverting the patch?
>
> I'm travelling at the moment and can't easily check it, but I would
> expect that my patch has no effect on the generated object code [...]
Michael, Arnd, thx for the replies. To you and everyone else that looked
into this: sorry for the trouble this caused.
The reporter's guess was wrong, as the reporter meanwhile confirmed in
the bugzilla ticket that the problem started to happen earlier.
Bagas, please be a bit more careful and don't blame a specific commit
unless it's was found by bisection, a revert through a lucky guess, a
statement from a developer, or something like that. In cases like this
it would have been better to sent the developers of said commit a quick
mail along the lines of "could you imagine that this change could lead
to the problem the reporter described". But even that might be too much
in a case like this, as too many of such false alarms and inquiries will
make developers start hating or ignoring regression tracking in general
or mails from you or me – and that is something that must be avoided, as
without help from developers regression tracking becomes a lot harder or
impossible.
Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat)
--
Everything you wanna know about Linux kernel regression tracking:
https://linux-regtracking.leemhuis.info/about/#tldr
If I did something stupid, please tell me, as explained on that page.
P.S.: Updating regzbot status, while at it:
#regzbot introduced: v6.1..v6.2
On 6/24/23 09:00, Linux regression tracking (Thorsten Leemhuis) wrote:
> On 24.06.23 11:29, Arnd Bergmann wrote:
>> On Sat, Jun 24, 2023, at 10:50, Michael Büsch wrote:
>>> On Sat, 24 Jun 2023 08:44:15 +0700
>>> Bagas Sanjaya <[email protected]> wrote:
>>>>> I suspect change introduced when addressing a compiler warning
>>>>> cased the error.
>>>>>
>>>>> https://patchwork.kernel.org/project/linux-wireless/patch/20230516183442.536589-1-arnd%40kernel.org/
>>>
>>> I doubt it.
>>> This patch affects the device initialization code. But the crash is in
>>> the transmit path.
>>> Can you please double check by manually reverting the patch?
>>
>> I'm travelling at the moment and can't easily check it, but I would
>> expect that my patch has no effect on the generated object code [...]
>
> Michael, Arnd, thx for the replies. To you and everyone else that looked
> into this: sorry for the trouble this caused.
>
> The reporter's guess was wrong, as the reporter meanwhile confirmed in
> the bugzilla ticket that the problem started to happen earlier.
>
> Bagas, please be a bit more careful and don't blame a specific commit
> unless it's was found by bisection, a revert through a lucky guess, a
> statement from a developer, or something like that. In cases like this
> it would have been better to sent the developers of said commit a quick
> mail along the lines of "could you imagine that this change could lead
> to the problem the reporter described". But even that might be too much
> in a case like this, as too many of such false alarms and inquiries will
> make developers start hating or ignoring regression tracking in general
> or mails from you or me – and that is something that must be avoided, as
> without help from developers regression tracking becomes a lot harder or
> impossible.
>
> Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat)
> --
> Everything you wanna know about Linux kernel regression tracking:
> https://linux-regtracking.leemhuis.info/about/#tldr
> If I did something stupid, please tell me, as explained on that page.
>
> P.S.: Updating regzbot status, while at it:
The OP definitely needs to bisect this problem. The only system that I have that
runs b43 is a PowerBook G4 with a ppc32 CPU. That box has successfully run b43
for every release up to 6.4.0-rc7. I have seen no problems in recent kernels.
Larry