dddd

dddd

[Helge Deller]

Dear Greg & stable team,

could you please queue up the patch below for the stable-6.7 kernel?
This is upstream commit:
	eba38cc7578bef94865341c73608bdf49193a51d

Thanks,
Helge


From eba38cc7578bef94865341c73608bdf49193a51d Mon Sep 17 00:00:00 2001
From: Helge Deller <deller@kernel.org>
Subject: [PATCH] bcachefs: Fix build on parisc by avoiding __multi3()

The gcc compiler on paric does support the __int128 type, although the
architecture does not have native 128-bit support.

The effect is, that the bcachefs u128_square() function will pull in the
libgcc __multi3() helper, which breaks the kernel build when bcachefs is
built as module since this function isn't currently exported in
arch/parisc/kernel/parisc_ksyms.c.
The build failure can be seen in the latest debian kernel build at:
https://buildd.debian.org/status/fetch.php?pkg=linux&arch=hppa&ver=6.7.1-1%7Eexp1&stamp=1706132569&raw=0

We prefer to not export that symbol, so fall back to the optional 64-bit
implementation provided by bcachefs and thus avoid usage of __multi3().

Signed-off-by: Helge Deller <deller@gmx.de>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

diff --git a/fs/bcachefs/mean_and_variance.h b/fs/bcachefs/mean_and_variance.h
index b2be565bb8f2..64df11ab422b 100644
--- a/fs/bcachefs/mean_and_variance.h
+++ b/fs/bcachefs/mean_and_variance.h
@@ -17,7 +17,7 @@
  * Rust and rustc has issues with u128.
  */
 
-#if defined(__SIZEOF_INT128__) && defined(__KERNEL__)
+#if defined(__SIZEOF_INT128__) && defined(__KERNEL__) && !defined(CONFIG_PARISC)
 
 typedef struct {
 	unsigned __int128 v;

Re: dddd

[Kent Overstreet]

On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
> Dear Greg & stable team,
> 
> could you please queue up the patch below for the stable-6.7 kernel?
> This is upstream commit:
> 	eba38cc7578bef94865341c73608bdf49193a51d
> 
> Thanks,
> Helge

I've already sent Greg a pull request with this patch - _twice_.

> 
> 
> From eba38cc7578bef94865341c73608bdf49193a51d Mon Sep 17 00:00:00 2001
> From: Helge Deller <deller@kernel.org>
> Subject: [PATCH] bcachefs: Fix build on parisc by avoiding __multi3()
> 
> The gcc compiler on paric does support the __int128 type, although the
> architecture does not have native 128-bit support.
> 
> The effect is, that the bcachefs u128_square() function will pull in the
> libgcc __multi3() helper, which breaks the kernel build when bcachefs is
> built as module since this function isn't currently exported in
> arch/parisc/kernel/parisc_ksyms.c.
> The build failure can be seen in the latest debian kernel build at:
> https://buildd.debian.org/status/fetch.php?pkg=linux&arch=hppa&ver=6.7.1-1%7Eexp1&stamp=1706132569&raw=0
> 
> We prefer to not export that symbol, so fall back to the optional 64-bit
> implementation provided by bcachefs and thus avoid usage of __multi3().
> 
> Signed-off-by: Helge Deller <deller@gmx.de>
> Cc: Kent Overstreet <kent.overstreet@linux.dev>
> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
> 
> diff --git a/fs/bcachefs/mean_and_variance.h b/fs/bcachefs/mean_and_variance.h
> index b2be565bb8f2..64df11ab422b 100644
> --- a/fs/bcachefs/mean_and_variance.h
> +++ b/fs/bcachefs/mean_and_variance.h
> @@ -17,7 +17,7 @@
>   * Rust and rustc has issues with u128.
>   */
>  
> -#if defined(__SIZEOF_INT128__) && defined(__KERNEL__)
> +#if defined(__SIZEOF_INT128__) && defined(__KERNEL__) && !defined(CONFIG_PARISC)
>  
>  typedef struct {
>  	unsigned __int128 v;

Re: bachefs stable patches [was dddd]

[Helge Deller]

(fixed email subject)
On 3/14/24 10:46, Kent Overstreet wrote:
> On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
>> Dear Greg & stable team,
>>
>> could you please queue up the patch below for the stable-6.7 kernel?
>> This is upstream commit:
>> 	eba38cc7578bef94865341c73608bdf49193a51d
>>
>> Thanks,
>> Helge
>
> I've already sent Greg a pull request with this patch - _twice_.

OIC.
You and Greg had some email exchange :-)

Greg, I'm seeing kernel build failures in debian with kernel 6.7.9
and the patch mentioned above isn't sufficient.

Would you please instead pull in those two upstream commits (in this order) to fix it:
44fd13a4c68e87953ccd827e764fa566ddcbbcf5  ("bcachefs: Fixes for rust bindgen")
eba38cc7578bef94865341c73608bdf49193a51d  ("bcachefs: Fix build on parisc by avoiding __multi3()")

Thanks,
Helge

Re: bachefs stable patches [was dddd]

[Kent Overstreet]

On Thu, Mar 14, 2024 at 01:57:51PM +0100, Helge Deller wrote:
> (fixed email subject)
> On 3/14/24 10:46, Kent Overstreet wrote:
> > On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
> > > Dear Greg & stable team,
> > > 
> > > could you please queue up the patch below for the stable-6.7 kernel?
> > > This is upstream commit:
> > > 	eba38cc7578bef94865341c73608bdf49193a51d
> > > 
> > > Thanks,
> > > Helge
> > 
> > I've already sent Greg a pull request with this patch - _twice_.
> 
> OIC.
> You and Greg had some email exchange :-)
> 
> Greg, I'm seeing kernel build failures in debian with kernel 6.7.9
> and the patch mentioned above isn't sufficient.
> 
> Would you please instead pull in those two upstream commits (in this order) to fix it:
> 44fd13a4c68e87953ccd827e764fa566ddcbbcf5  ("bcachefs: Fixes for rust bindgen")

You'll have to explain what this patch fixes, it shouldn't be doing
anything when building in the kernel (yet; it will when we've pulled our
Rust code into the kernel).

Re: bachefs stable patches [was dddd]

[Helge Deller]

On 3/14/24 20:08, Kent Overstreet wrote:
> On Thu, Mar 14, 2024 at 01:57:51PM +0100, Helge Deller wrote:
>> (fixed email subject)
>> On 3/14/24 10:46, Kent Overstreet wrote:
>>> On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
>>>> Dear Greg & stable team,
>>>>
>>>> could you please queue up the patch below for the stable-6.7 kernel?
>>>> This is upstream commit:
>>>> 	eba38cc7578bef94865341c73608bdf49193a51d
>>>>
>>>> Thanks,
>>>> Helge
>>>
>>> I've already sent Greg a pull request with this patch - _twice_.
>>
>> OIC.
>> You and Greg had some email exchange :-)
>>
>> Greg, I'm seeing kernel build failures in debian with kernel 6.7.9
>> and the patch mentioned above isn't sufficient.
>>
>> Would you please instead pull in those two upstream commits (in this order) to fix it:
>> 44fd13a4c68e87953ccd827e764fa566ddcbbcf5  ("bcachefs: Fixes for rust bindgen")
>
> You'll have to explain what this patch fixes, it shouldn't be doing
> anything when building in the kernel (yet; it will when we've pulled our
> Rust code into the kernel).

Right. It doesn't actually do anything in the kernel (which is good), but
it's logically before patch eba38cc7578bef94865341c73608bdf49193a51d
and is required so that eba38cc7578bef94865341c73608bdf49193a51d cleanly
applies.

Helge

Re: bachefs stable patches [was dddd]

[Kent Overstreet]

On Thu, Mar 14, 2024 at 10:34:59PM +0100, Helge Deller wrote:
> On 3/14/24 20:08, Kent Overstreet wrote:
> > On Thu, Mar 14, 2024 at 01:57:51PM +0100, Helge Deller wrote:
> > > (fixed email subject)
> > > On 3/14/24 10:46, Kent Overstreet wrote:
> > > > On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
> > > > > Dear Greg & stable team,
> > > > > 
> > > > > could you please queue up the patch below for the stable-6.7 kernel?
> > > > > This is upstream commit:
> > > > > 	eba38cc7578bef94865341c73608bdf49193a51d
> > > > > 
> > > > > Thanks,
> > > > > Helge
> > > > 
> > > > I've already sent Greg a pull request with this patch - _twice_.
> > > 
> > > OIC.
> > > You and Greg had some email exchange :-)
> > > 
> > > Greg, I'm seeing kernel build failures in debian with kernel 6.7.9
> > > and the patch mentioned above isn't sufficient.
> > > 
> > > Would you please instead pull in those two upstream commits (in this order) to fix it:
> > > 44fd13a4c68e87953ccd827e764fa566ddcbbcf5  ("bcachefs: Fixes for rust bindgen")
> > 
> > You'll have to explain what this patch fixes, it shouldn't be doing
> > anything when building in the kernel (yet; it will when we've pulled our
> > Rust code into the kernel).
> 
> Right. It doesn't actually do anything in the kernel (which is good), but
> it's logically before patch eba38cc7578bef94865341c73608bdf49193a51d
> and is required so that eba38cc7578bef94865341c73608bdf49193a51d cleanly
> applies.

yeah I just fixed the merge conflict

Re: bachefs stable patches [was dddd]

[Helge Deller]

On 3/14/24 23:19, Kent Overstreet wrote:
> On Thu, Mar 14, 2024 at 10:34:59PM +0100, Helge Deller wrote:
>> On 3/14/24 20:08, Kent Overstreet wrote:
>>> On Thu, Mar 14, 2024 at 01:57:51PM +0100, Helge Deller wrote:
>>>> (fixed email subject)
>>>> On 3/14/24 10:46, Kent Overstreet wrote:
>>>>> On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
>>>>>> Dear Greg & stable team,
>>>>>>
>>>>>> could you please queue up the patch below for the stable-6.7 kernel?
>>>>>> This is upstream commit:
>>>>>> 	eba38cc7578bef94865341c73608bdf49193a51d
>>>>>>
>>>>>> Thanks,
>>>>>> Helge
>>>>>
>>>>> I've already sent Greg a pull request with this patch - _twice_.
>>>>
>>>> OIC.
>>>> You and Greg had some email exchange :-)
>>>>
>>>> Greg, I'm seeing kernel build failures in debian with kernel 6.7.9
>>>> and the patch mentioned above isn't sufficient.
>>>>
>>>> Would you please instead pull in those two upstream commits (in this order) to fix it:
>>>> 44fd13a4c68e87953ccd827e764fa566ddcbbcf5  ("bcachefs: Fixes for rust bindgen")
>>>
>>> You'll have to explain what this patch fixes, it shouldn't be doing
>>> anything when building in the kernel (yet; it will when we've pulled our
>>> Rust code into the kernel).
>>
>> Right. It doesn't actually do anything in the kernel (which is good), but
>> it's logically before patch eba38cc7578bef94865341c73608bdf49193a51d
>> and is required so that eba38cc7578bef94865341c73608bdf49193a51d cleanly
>> applies.
>
> yeah I just fixed the merge conflict

Where did you fixed it?
Those two patches are already upstream, and ideally should go downstream as-is...

Helge

Re: bachefs stable patches [was dddd]

[Kent Overstreet]

On Thu, Mar 14, 2024 at 11:25:28PM +0100, Helge Deller wrote:
> On 3/14/24 23:19, Kent Overstreet wrote:
> > On Thu, Mar 14, 2024 at 10:34:59PM +0100, Helge Deller wrote:
> > > On 3/14/24 20:08, Kent Overstreet wrote:
> > > > On Thu, Mar 14, 2024 at 01:57:51PM +0100, Helge Deller wrote:
> > > > > (fixed email subject)
> > > > > On 3/14/24 10:46, Kent Overstreet wrote:
> > > > > > On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
> > > > > > > Dear Greg & stable team,
> > > > > > > 
> > > > > > > could you please queue up the patch below for the stable-6.7 kernel?
> > > > > > > This is upstream commit:
> > > > > > > 	eba38cc7578bef94865341c73608bdf49193a51d
> > > > > > > 
> > > > > > > Thanks,
> > > > > > > Helge
> > > > > > 
> > > > > > I've already sent Greg a pull request with this patch - _twice_.
> > > > > 
> > > > > OIC.
> > > > > You and Greg had some email exchange :-)
> > > > > 
> > > > > Greg, I'm seeing kernel build failures in debian with kernel 6.7.9
> > > > > and the patch mentioned above isn't sufficient.
> > > > > 
> > > > > Would you please instead pull in those two upstream commits (in this order) to fix it:
> > > > > 44fd13a4c68e87953ccd827e764fa566ddcbbcf5  ("bcachefs: Fixes for rust bindgen")
> > > > 
> > > > You'll have to explain what this patch fixes, it shouldn't be doing
> > > > anything when building in the kernel (yet; it will when we've pulled our
> > > > Rust code into the kernel).
> > > 
> > > Right. It doesn't actually do anything in the kernel (which is good), but
> > > it's logically before patch eba38cc7578bef94865341c73608bdf49193a51d
> > > and is required so that eba38cc7578bef94865341c73608bdf49193a51d cleanly
> > > applies.
> > 
> > yeah I just fixed the merge conflict
> 
> Where did you fixed it?
> Those two patches are already upstream, and ideally should go downstream as-is...

https://evilpiepirate.org/git/bcachefs.git/log/?h=bcachefs-for-v6.7

Re: dddd

[Sasha Levin]

On Thu, Mar 14, 2024 at 05:46:35AM -0400, Kent Overstreet wrote:
>On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
>> Dear Greg & stable team,
>>
>> could you please queue up the patch below for the stable-6.7 kernel?
>> This is upstream commit:
>> 	eba38cc7578bef94865341c73608bdf49193a51d
>>
>> Thanks,
>> Helge
>
>I've already sent Greg a pull request with this patch - _twice_.

I'll point out, again, that if you read the docs it clearly points out
that pull requests aren't a way to submit patches into stable.

-- 
Thanks,
Sasha

Re: dddd

[Kent Overstreet]

On Thu, Mar 14, 2024 at 06:38:19PM -0400, Sasha Levin wrote:
> On Thu, Mar 14, 2024 at 05:46:35AM -0400, Kent Overstreet wrote:
> > On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
> > > Dear Greg & stable team,
> > > 
> > > could you please queue up the patch below for the stable-6.7 kernel?
> > > This is upstream commit:
> > > 	eba38cc7578bef94865341c73608bdf49193a51d
> > > 
> > > Thanks,
> > > Helge
> > 
> > I've already sent Greg a pull request with this patch - _twice_.
> 
> I'll point out, again, that if you read the docs it clearly points out
> that pull requests aren't a way to submit patches into stable.

Sasha, Greg and I already discussed and agreed that this would be the
plan for fs/bcachefs/.

Re: dddd

[Greg Kroah-Hartman]

On Thu, Mar 14, 2024 at 07:02:46PM -0400, Kent Overstreet wrote:
> On Thu, Mar 14, 2024 at 06:38:19PM -0400, Sasha Levin wrote:
> > On Thu, Mar 14, 2024 at 05:46:35AM -0400, Kent Overstreet wrote:
> > > On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
> > > > Dear Greg & stable team,
> > > > 
> > > > could you please queue up the patch below for the stable-6.7 kernel?
> > > > This is upstream commit:
> > > > 	eba38cc7578bef94865341c73608bdf49193a51d
> > > > 
> > > > Thanks,
> > > > Helge
> > > 
> > > I've already sent Greg a pull request with this patch - _twice_.
> > 
> > I'll point out, again, that if you read the docs it clearly points out
> > that pull requests aren't a way to submit patches into stable.
> 
> Sasha, Greg and I already discussed and agreed that this would be the
> plan for fs/bcachefs/.

I agreed?  I said that I would turn a pull request into individual
patches, but that it's extra work for me to do so so that it usually
will end up at the end of my work queue.  I'll get to this when I get
back from vacation and catch up with other stuff in a few weeks, thanks.

greg k-h

Re: dddd

[Kent Overstreet]

On Fri, Mar 15, 2024 at 07:27:23PM +0100, Greg Kroah-Hartman wrote:
> On Thu, Mar 14, 2024 at 07:02:46PM -0400, Kent Overstreet wrote:
> > On Thu, Mar 14, 2024 at 06:38:19PM -0400, Sasha Levin wrote:
> > > On Thu, Mar 14, 2024 at 05:46:35AM -0400, Kent Overstreet wrote:
> > > > On Thu, Mar 14, 2024 at 10:41:12AM +0100, Helge Deller wrote:
> > > > > Dear Greg & stable team,
> > > > > 
> > > > > could you please queue up the patch below for the stable-6.7 kernel?
> > > > > This is upstream commit:
> > > > > 	eba38cc7578bef94865341c73608bdf49193a51d
> > > > > 
> > > > > Thanks,
> > > > > Helge
> > > > 
> > > > I've already sent Greg a pull request with this patch - _twice_.
> > > 
> > > I'll point out, again, that if you read the docs it clearly points out
> > > that pull requests aren't a way to submit patches into stable.
> > 
> > Sasha, Greg and I already discussed and agreed that this would be the
> > plan for fs/bcachefs/.
> 
> I agreed?  I said that I would turn a pull request into individual
> patches, but that it's extra work for me to do so so that it usually
> will end up at the end of my work queue.  I'll get to this when I get
> back from vacation and catch up with other stuff in a few weeks, thanks.

No, you never said anything about turning a pull request into individual
patches; that was something I found out after you repeatedly mangled my
pull requests.

But how the pull requests get applied is entirely beside the point right
now, because apparently you agreed to a process without making any
provisions for having that continue while you were on vacation?

So now we've got a user data eating bug out there, with a fix that's
been sent to you, and you guys are just sitting on your hands.

You guys need to get your shit together - I've already let people know
that 6.7 is not safe to use and they need to immediately upgrade to 6.8,
but unfortunately not everyone will see that.

This is going to cause people reall pain.