Re: [PATCH v2] Fix linux-starfive build under sid

["'Jan Kiszka' via isar-users" <isar-users@googlegroups.com>]

On 04.10.24 12:49, Uladzimir Bely wrote:
> On Fri, 2024-10-04 at 12:45 +0200, Jan Kiszka wrote:
>> On 04.10.24 12:37, Uladzimir Bely wrote:
>>> On Fri, 2024-10-04 at 12:27 +0200, Jan Kiszka wrote:
>>>> On 04.10.24 12:22, 'Jan Kiszka' via isar-users wrote:
>>>>> On 04.10.24 11:43, Uladzimir Bely wrote:
>>>>>> On Fri, 2024-10-04 at 10:38 +0200, Jan Kiszka wrote:
>>>>>>> On 04.10.24 10:17, Uladzimir Bely wrote:
>>>>>>>> The patch is taken from the linux kernel upstream >= 6.8.
>>>>>>>>
>>>>>>>> Signed-off-by: Uladzimir Bely <ubely@ilbers.de>
>>>>>>>> ---
>>>>>>>>  ...oc-arguments-order-in-btrfs_ioctl_se.patch | 43
>>>>>>>> +++++++++++++++++++
>>>>>>>>  .../linux/linux-starfive_6.6-visionfive2.bb   |  1 +
>>>>>>>>  2 files changed, 44 insertions(+)
>>>>>>>>  create mode 100644 meta-isar/recipes-
>>>>>>>> kernel/linux/files/0001-
>>>>>>>> btrfs-fix-kvcalloc-arguments-order-in-
>>>>>>>> btrfs_ioctl_se.patch
>>>>>>>>
>>>>>>>> Changes since v1 (Fix linux build under sid and trixie):
>>>>>>>>
>>>>>>>> - Since we don't build linux-mainline/linux-phy kernels
>>>>>>>> for
>>>>>>>> trixie/sid
>>>>>>>> apply kernel patch only for linux-starfive_6.6-
>>>>>>>> visionfive2
>>>>>>>> target.
>>>>>>>>
>>>>>>>> I additionally checked [1] for existing branches and can
>>>>>>>> conclude:
>>>>>>>>  - recent 6.6-based branch "JH7110_VisionFive2_upstream"
>>>>>>>> is
>>>>>>>> still
>>>>>>>> affected and requires the patch
>>>>>>>>  - recent "starfive-tech/visionfive" is 6.10-based
>>>>>>>> already
>>>>>>>> includes
>>>>>>>> the fix, but using it leads to different build errors.
>>>>>>>> Also,
>>>>>>>> I'm
>>>>>>>> not
>>>>>>>> sure if it properly supports "VisionFive2" board.
>>>>>>>>
>>>>>>>> So, for now, the simplest way to fix build error we have
>>>>>>>> in
>>>>>>>> CI is
>>>>>>>> to
>>>>>>>> just apply upstream patch.
>>>>>>>>
>>>>>>>> [1] https://github.com/starfive-tech/linux
>>>>>>>>
>>>>>>>> diff --git a/meta-isar/recipes-kernel/linux/files/0001-
>>>>>>>> btrfs-
>>>>>>>> fix-
>>>>>>>> kvcalloc-arguments-order-in-btrfs_ioctl_se.patch b/meta-
>>>>>>>> isar/recipes-kernel/linux/files/0001-btrfs-fix-kvcalloc-
>>>>>>>> arguments-
>>>>>>>> order-in-btrfs_ioctl_se.patch
>>>>>>>> new file mode 100644
>>>>>>>> index 00000000..afe09eab
>>>>>>>> --- /dev/null
>>>>>>>> +++ b/meta-isar/recipes-kernel/linux/files/0001-btrfs-
>>>>>>>> fix-
>>>>>>>> kvcalloc-
>>>>>>>> arguments-order-in-btrfs_ioctl_se.patch
>>>>>>>> @@ -0,0 +1,43 @@
>>>>>>>> +From 6ff09b6b8c2fb6b3edda4ffaa173153a40653067 Mon Sep 17
>>>>>>>> 00:00:00
>>>>>>>> 2001
>>>>>>>> +From: Dmitry Antipov <dmantipov@yandex.ru>
>>>>>>>> +Date: Thu, 21 Dec 2023 11:47:45 +0300
>>>>>>>> +Subject: [PATCH] btrfs: fix kvcalloc() arguments order
>>>>>>>> in
>>>>>>>> btrfs_ioctl_send()
>>>>>>>> +
>>>>>>>> +When compiling with gcc version 14.0.0 20231220
>>>>>>>> (experimental)
>>>>>>>> +and W=1, I've noticed the following warning:
>>>>>>>> +
>>>>>>>> +fs/btrfs/send.c: In function 'btrfs_ioctl_send':
>>>>>>>> +fs/btrfs/send.c:8208:44: warning: 'kvcalloc' sizes
>>>>>>>> specified
>>>>>>>> with
>>>>>>>> 'sizeof'
>>>>>>>> +in the earlier argument and not in the later argument [-
>>>>>>>> Wcalloc-
>>>>>>>> transposed-args]
>>>>>>>> + 8208 |         sctx->clone_roots =
>>>>>>>> kvcalloc(sizeof(*sctx-
>>>>>>>>> clone_roots),
>>>>>>>> +      |                                            ^
>>>>>>>> +
>>>>>>>> +Since 'n' and 'size' arguments of 'kvcalloc()' are
>>>>>>>> multiplied to
>>>>>>>> +calculate the final size, their actual order doesn't
>>>>>>>> affect
>>>>>>>> the
>>>>>>>> result
>>>>>>>> +and so this is not a bug. But it's still worth to fix
>>>>>>>> it.
>>>>>>>> +
>>>>>>>> +Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
>>>>>>>> +Reviewed-by: David Sterba <dsterba@suse.com>
>>>>>>>> +Signed-off-by: David Sterba <dsterba@suse.com>
>>>>>>>> +---
>>>>>>>> + fs/btrfs/send.c | 4 ++--
>>>>>>>> + 1 file changed, 2 insertions(+), 2 deletions(-)
>>>>>>>> +
>>>>>>>> +diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c
>>>>>>>> +index 4e36550618e5..2d7519a6ce72 100644
>>>>>>>> +--- a/fs/btrfs/send.c
>>>>>>>> ++++ b/fs/btrfs/send.c
>>>>>>>> +@@ -8205,8 +8205,8 @@ long btrfs_ioctl_send(struct inode
>>>>>>>> *inode,
>>>>>>>> struct btrfs_ioctl_send_args *arg)
>>>>>>>> + 		goto out;
>>>>>>>> + 	}
>>>>>>>> + 
>>>>>>>> +-	sctx->clone_roots = kvcalloc(sizeof(*sctx-
>>>>>>>>> clone_roots),
>>>>>>>> +-				     arg-
>>>>>>>>> clone_sources_count + 1,
>>>>>>>> ++	sctx->clone_roots = kvcalloc(arg-
>>>>>>>>> clone_sources_count + 1,
>>>>>>>> ++				     sizeof(*sctx-
>>>>>>>>> clone_roots),
>>>>>>>> + 				     GFP_KERNEL);
>>>>>>>> + 	if (!sctx->clone_roots) {
>>>>>>>> + 		ret = -ENOMEM;
>>>>>>>> +-- 
>>>>>>>> +2.45.2
>>>>>>>> +
>>>>>>>> diff --git a/meta-isar/recipes-kernel/linux/linux-
>>>>>>>> starfive_6.6-
>>>>>>>> visionfive2.bb b/meta-isar/recipes-kernel/linux/linux-
>>>>>>>> starfive_6.6-
>>>>>>>> visionfive2.bb
>>>>>>>> index f4dd5c9a..a267241f 100644
>>>>>>>> --- a/meta-isar/recipes-kernel/linux/linux-starfive_6.6-
>>>>>>>> visionfive2.bb
>>>>>>>> +++ b/meta-isar/recipes-kernel/linux/linux-starfive_6.6-
>>>>>>>> visionfive2.bb
>>>>>>>> @@ -8,6 +8,7 @@ require recipes-kernel/linux/linux-
>>>>>>>> custom.inc
>>>>>>>>  SRC_URI += " \
>>>>>>>>     
>>>>>>>> https://github.com/starfive-tech/linux/archive/${SRCREV}.tar.gz;downloadfilename=linux-starfive-${SRCREV}.tar.gz
>>>>>>>>  \
>>>>>>>>     
>>>>>>>> file://0001-riscv-efistub-Ensure-GP-relative-addressing-is-not-u.patch
>>>>>>>>  \
>>>>>>>> +   
>>>>>>>> file://0001-btrfs-fix-kvcalloc-arguments-order-in-btrfs_ioctl_se.patch
>>>>>>>>  \
>>>>>>>>      file://starfive2_extra.cfg"
>>>>>>>>  SRCREV = "9fe004eaf1aa5b23bd5d03b4cfe9c3858bd884c4"
>>>>>>>>  SRC_URI[sha256sum] =
>>>>>>>> "9eaf7659aa57e2c5b399b7b33076f1376ec43ef343680e0a57e0a2a9
>>>>>>>> bef6
>>>>>>>> c0ae"
>>>>>>>
>>>>>>> NACK, update the kernel. I think I already told you. May
>>>>>>> even
>>>>>>> obsolete
>>>>>>> the other patch.
>>>>>>
>>>>>> Yes, I tried this (see the text above the diff starts).
>>>>>> Updated
>>>>>> 6.6-
>>>>>> based branch still requires both patches.
>>>>>>
>>>>>
>>>>> https://github.com/starfive-tech/linux/commits/starfive-6.6.31-dubhe
>>>>>
>>>>> As I told you.
>>>>>
>>>>> I'm sending the proper patch in a minute.
>>>>>
>>>>
>>>> Sorry, seems that's the wrong tree, and
>>>> JH7110_VisionFive2_6.6.y_devel
>>>> was not updated properly.
>>>>
>>>> But there is no way to build against sid except for using an
>>>> older
>>>> snapshot. How did you test this?
>>>>
>>>> Jan
>>>>
>>>
>>> I tested few branches:
>>> - JH7110_VisionFive2_upstream - this seems to be a proper tree (at
>>> least, it includes starfive_visionfive2_defconfig, and appropriate
>>> devicetree files look complete). But this still has now a fix and
>>> requires both patches
>>
>> Yes, it is. I've created
>> https://github.com/starfive-tech/linux/issues/140 to resolve the
>> problem
>> "upstream" eventually.
>>
>>> - starfive-tech/visionfive - this is based on linux-6.10, includes
>>> both patches. But there are different build errors here and no
>>> starfive_visionfive2_defconfig present. Also, board devicetree
>>> files
>>> has less stuff.
>>
>> That's version 1 of the VisionFive, a different board.
>>
>>> - starfive-6.6.31-dubhe - this includes the patch (but not the
>>> other
>>> one). But there is also no proper defconfig and board deficetrees
>>> are
>>> not looking full enough.
>>
>> Even a different SoC that may lack patches we need. I was misled.
>>
>>>
>>> Some tests were done in one of our CI instances where we already
>>> have
>>> an updated version of debootstrap installed. Manual check of the
>>> build
>>> can be also done under kas-container shell, but some additional
>>> steps
>>> (e.g., adding bookworm-backports to source list and installing
>>> "debootstrap/bookworm-backports") are required before running
>>> bitbake.
>>>
>>
>> You must be natively building - I don't have that much time.
>> Crossbuild
>> is borken once again, but the snapshot we also use in isar-cip-core
>> is
>> fine (but not affected by the compiler update), see my kas menu
>> patch.
>>
> 
> Yes, I was building it natively for tests (as well as CI does).
> 
> Trixie/sid cross builds are currently broken (not only on riscv64).
> That's another issue to solve (or, maybe, wait for upstream?).
> 

Use snapshots for crossbuilding. As I wrote, native building takes way
too much time to be used in practice.

BTW, we can bump to the latest revision of
JH7110_VisionFive2_6.6.y_devel and drop the other patch at this chance.
Just noticed that 6.6.20, their current head, actually contains it.

Jan

-- 
Siemens AG, Technology
Linux Expert Center

-- 
You received this message because you are subscribed to the Google Groups "isar-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isar-users+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/isar-users/bd380709-be0d-420a-b49d-46dcfc59e0c1%40siemens.com.