From: Jeffrey Hugo <quic_jhugo@quicinc.com>
To: Greg KH <greg@kroah.com>, Stephen Rothwell <sfr@canb.auug.org.au>,
"Oded Gabbay" <ogabbay@kernel.org>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
DRI <dri-devel@lists.freedesktop.org>,
Pranjal Ramajor Asha Kanojiya <quic_pkanojiy@quicinc.com>,
"Linux Next Mailing List" <linux-next@vger.kernel.org>,
Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>,
Dave Airlie <airlied@redhat.com>,
Manivannan Sadhasivam <mani@kernel.org>
Subject: Re: linux-next: build failure after merge of the driver-core tree
Date: Tue, 11 Apr 2023 12:37:23 -0600 [thread overview]
Message-ID: <086e08a2-13b3-870c-4b17-1fdc9d56d551@quicinc.com> (raw)
In-Reply-To: <ZDWlIuRHYPP1DeYi@phenom.ffwll.local>
On 4/11/2023 12:21 PM, Daniel Vetter wrote:
> On Tue, Apr 11, 2023 at 11:18:29AM -0600, Jeffrey Hugo wrote:
>> On 4/11/2023 10:31 AM, Daniel Vetter wrote:
>>> On Tue, Apr 11, 2023 at 09:29:27AM -0600, Jeffrey Hugo wrote:
>>>> On 4/11/2023 9:26 AM, Jeffrey Hugo wrote:
>>>>> On 4/11/2023 9:13 AM, Greg KH wrote:
>>>>>> On Tue, Apr 11, 2023 at 09:08:39AM -0600, Jeffrey Hugo wrote:
>>>>>>> On 4/11/2023 9:01 AM, Daniel Vetter wrote:
>>>>>>>> On Tue, Apr 11, 2023 at 12:40:28PM +0200, Greg KH wrote:
>>>>>>>>> On Tue, Apr 11, 2023 at 11:55:20AM +0200, Daniel Vetter wrote:
>>>>>>>>>> On Tue, Apr 11, 2023 at 02:38:12PM +1000, Stephen Rothwell wrote:
>>>>>>>>>>> Hi all,
>>>>>>>>>>>
>>>>>>>>>>> After merging the driver-core tree, today's linux-next build (x86_64
>>>>>>>>>>> allmodconfig) failed like this:
>>>>>>>>>>>
>>>>>>>>>>> In file included from include/linux/linkage.h:7,
>>>>>>>>>>> from include/linux/kernel.h:17,
>>>>>>>>>>> from drivers/accel/qaic/mhi_qaic_ctrl.c:4:
>>>>>>>>>>> drivers/accel/qaic/mhi_qaic_ctrl.c: In function
>>>>>>>>>>> 'mhi_qaic_ctrl_init':
>>>>>>>>>>> include/linux/export.h:27:22: error: passing
>>>>>>>>>>> argument 1 of 'class_create' from incompatible
>>>>>>>>>>> pointer type
>>>>>>>>>>> [-Werror=incompatible-pointer-types]
>>>>>>>>>>> 27 | #define THIS_MODULE (&__this_module)
>>>>>>>>>>> | ~^~~~~~~~~~~~~~~
>>>>>>>>>>> | |
>>>>>>>>>>> | struct module *
>>>>>>>>>>> drivers/accel/qaic/mhi_qaic_ctrl.c:544:38: note:
>>>>>>>>>>> in expansion of macro 'THIS_MODULE'
>>>>>>>>>>> 544 | mqc_dev_class =
>>>>>>>>>>> class_create(THIS_MODULE,
>>>>>>>>>>> MHI_QAIC_CTRL_DRIVER_NAME);
>>>>>>>>>>> | ^~~~~~~~~~~
>>>>>>>>>>> In file included from include/linux/device.h:31,
>>>>>>>>>>> from include/linux/mhi.h:9,
>>>>>>>>>>> from drivers/accel/qaic/mhi_qaic_ctrl.c:5:
>>>>>>>>>>> include/linux/device/class.h:229:54: note:
>>>>>>>>>>> expected 'const char *' but argument is of type
>>>>>>>>>>> 'struct module *'
>>>>>>>>>>> 229 | struct class * __must_check
>>>>>>>>>>> class_create(const char *name);
>>>>>>>>>>> | ~~~~~~~~~~~~^~~~
>>>>>>>>>>> drivers/accel/qaic/mhi_qaic_ctrl.c:544:25:
>>>>>>>>>>> error: too many arguments to function
>>>>>>>>>>> 'class_create'
>>>>>>>>>>> 544 | mqc_dev_class =
>>>>>>>>>>> class_create(THIS_MODULE,
>>>>>>>>>>> MHI_QAIC_CTRL_DRIVER_NAME);
>>>>>>>>>>> | ^~~~~~~~~~~~
>>>>>>>>>>> include/linux/device/class.h:229:29: note: declared here
>>>>>>>>>>> 229 | struct class * __must_check
>>>>>>>>>>> class_create(const char *name);
>>>>>>>>>>> | ^~~~~~~~~~~~
>>>>>>>>>>>
>>>>>>>>>>> Caused by commit
>>>>>>>>>>>
>>>>>>>>>>> 1aaba11da9aa ("driver core: class: remove
>>>>>>>>>>> module * from class_create()")
>>>>>>>>>>>
>>>>>>>>>>> interacting with commit
>>>>>>>>>>>
>>>>>>>>>>> 566fc96198b4 ("accel/qaic: Add mhi_qaic_cntl")
>>>>>>>>>>>
>>>>>>>>>>> from the drm tree.
>>>>>>>>>>>
>>>>>>>>>>> I have applied the following merge fix patch for today.
>>>>>>>>>>>
>>>>>>>>>>> From: Stephen Rothwell <sfr@canb.auug.org.au>
>>>>>>>>>>> Date: Tue, 11 Apr 2023 14:16:57 +1000
>>>>>>>>>>> Subject: [PATCH] fixup for "driver core: class:
>>>>>>>>>>> remove module * from class_create()"
>>>>>>>>>>>
>>>>>>>>>>> interacting with "accel/qaic: Add mhi_qaic_cntl"
>>>>>>>>>>>
>>>>>>>>>>> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
>>>>>>>>>>
>>>>>>>>>> Thanks for the fixup. Since Dave is out I've made a
>>>>>>>>>> note about this in my
>>>>>>>>>> handover mail so it won't get lost in the drm-next
>>>>>>>>>> merge window pull. I
>>>>>>>>>> don't think we need any other coordination than
>>>>>>>>>> mention it in each pull to
>>>>>>>>>> Linus, topic tree seems overkill for this. Plus there's no way I can
>>>>>>>>>> untangle the drm tree anyway :-).
>>>>>>>>>
>>>>>>>>> Want me to submit a patch for the drm tree that moves this to use
>>>>>>>>> class_register() instead, which will make the
>>>>>>>>> merge/build issue go away
>>>>>>>>> for you? That's my long-term goal here anyway, so converting this new
>>>>>>>>> code to this api today would be something I have to do eventually :)
>>>>>>>>
>>>>>>>> We kinda closed drm-next for feature work mostly already (just pulling
>>>>>>>> stuff in from subtrees), so won't really help for this merge window.
>>>>>>>>
>>>>>>>> For everything else I think this is up to Oded, I had no
>>>>>>>> idea qaic needed
>>>>>>>> it's entire own dev class and I don't want to dig into this
>>>>>>>> for the risk I
>>>>>>>> might freak out :-)
>>>>>>>>
>>>>>>>> Adding Oded.
>>>>>>>>
>>>>>>>> Cheers, Daniel
>>>>>>>
>>>>>>> Sorry for the mess.
>>>>>>>
>>>>>>> I made a note to update to class_register() once my drm-misc access is
>>>>>>> sorted out. Looks like we'll address the conflict in the merge
>>>>>>> window, and
>>>>>>> catch the update to the new API in the following release.
>>>>>>
>>>>>> Wait, I think the large question is, "why does this need a separate
>>>>>> class"? Why are you not using the accel char device and class? That is
>>>>>> what everything under accel/ should be using, otherwise why put it in
>>>>>> there?
>>>>>>
>>>>>> And what exactly are you using that class for? Just device nodes? If
>>>>>> so, how many?
>>>>>>
>>>>>> thanks,
>>>>>>
>>>>>> greg k-h
>>>>>
>>>>>
>>>>> Remember MHI_UCI that then evolved into the WWAN subsystem? I pointed
>>>>> out at the time that AIC100/QAIC would need the same functionality.
>>>>> You/Jakub told myself/Mani/Loic that a combined implementation is not
>>>>> acceptable, and every area needs to implement their own version of
>>>>> MHI_UCI.
>>>>>
>>>>> We took the WWAN subsystem and simplified it to meet our needs.
>>>>>
>>>>> The functionality is QAIC specific, so wedging it into the Accel node
>>>>> seems to be a poor fit as it would subject Habana and iVPU to the same.
>>>>
>>>> Also, I forgot to mention. QAIC is sharing userspace components with WWAN,
>>>> so we really cannot diverge from what WWAN has done and define a new API
>>>> through the Accel node.
>>>
>>> So there is an accel/drm_device in the qaic driver, but there's also this
>>> different class thing, which I don't get.
>>>
>>> And yeah if that's an entirely orthogonal thing then I guess that should
>>> be in a different driver/subsystem, all supported with the aux bus to
>>> multiplex the underlying device.
>>>
>>> I haven't found any explanation for what MHI is (or any of the other
>>> acrynoms), so I'm entirely lost.
>>
>> MHI is documented at Documentation/mhi/
>> It is also referenced in the QAIC documentation - Documentation/accel/qaic/
>>
>> It stands for "Modem Host Interface" (arguably a bad name now, but you can
>> guess where it came from). It is a Qualcomm hardware block and associated
>> software protocol that provides logical channels over a hardware link. Most
>> commonly used for PCIe.
>>
>> Pretty much any modern Qualcomm PCIe device implements it. 4G modems, 5G
>> modems, Wifi adapters, AIC100, etc. Instead of talking "PCIe", the host
>> talks "MHI" to the devices in most cases.
>>
>> The core implementation for MHI exists in drivers/bus/mhi
>>
>> MHI_UCI is the MHI Userspace Character Interface. It looked like most buses
>> (eg USB) provide some direct device access to userspace. MHI_UCI was
>> formulated along those same lines - provide direct userspace access to a
>> whitelist of channels. Qualcomm provides some fairly extensive userspace
>> utilities, and various communities have developed open source alternatives
>> using this mechanism.
>>
>> MHI_UCI was proposed to the community as the common driver (misc device) for
>> all of the MHI devices. The Net folks came along, saw that it was used for
>> 4G/5G modems (Wireless Wide Area Network devices or WWAN) and decided that
>> they would not tolerate a common implementation. They NACK'd MHI_UCI and
>> required that a WWAN specific subsystem be developed which would only
>> service WWAN devices. The Net folks decreed that other subsystems which
>> needed the same functionality need to have their own copy of the
>> implementation.
>>
>> QAIC devices expose Sahara (a boot time protocol) which has an existing
>> userspace that is also used with Modems, although it looks like WWAN doesn't
>> currently support those generations of products today. QAIC devices also
>> support DIAG, which is currently supported in WWAN. The intent was to add
>> the QAIC support for DIAG at a later time since it is not required for the
>> bare minimum viable driver.
>>
>> So, QAIC devices support the same services, would use the same userspace,
>> but can't use a common implementation because Jakub(net) doesn't want to
>> share and convinced Greg to go along. I'm not interested in pushing a cross
>> tree fight (arguably already did that with MHI_UCI). If neither Greg nor
>> Net will accept a common implementation that accelerators can use (QAIC),
>> then the only place I can fit this is in the Accel area.
>>
>> Using aux bus seems to make little difference if QAIC is the only consumer
>> of this. I'm willing to refactor the implementation with some feedback and
>> guidence, but the uAPI seems set in stone due to the existing userspace and
>> WWAN (char devs with open/close/read/write/poll).
>
> Ok, so MHI _is_ the bus. Thanks for the explainer, I should have searched
> a bit more in Documentation/
>
>> What would make you less unhappy?
>
> The MHI generic userspace driver interface needs to be in drivers/bus/mhi,
> not in a random driver. I think we should revert 566fc96198b4
> ("accel/qaic: Add mhi_qaic_cntl") and re-land that through Greg's tree (or
> wherever mhi patches go to). This of course assuming that the accel
> userspace on top of the accel/drm_device does work stand-alone, and it's
> just the tooling and other userspace that needs MHI_UCI. If we end with a
> non-functional stack due to that, then I guess the entire driver is a bit
> up for questions, because at least the accel runtime is supposed to just
> run on top of the accel devnode and nothing else. Otherwise container
> stuff gets really bad, among a lot of other things.
>
Looping in the MHI maintainer for your proposal.
The accel userspace can work without MHI_UCI.
The revert will be non-trivial so I'll look at posting that tomorrow.
-Jeff
next prev parent reply other threads:[~2023-04-11 18:37 UTC|newest]
Thread overview: 181+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-04-11 4:38 linux-next: build failure after merge of the driver-core tree Stephen Rothwell
2023-04-11 9:55 ` Daniel Vetter
2023-04-11 10:40 ` Greg KH
2023-04-11 15:01 ` Daniel Vetter
2023-04-11 15:08 ` Jeffrey Hugo
2023-04-11 15:13 ` Greg KH
2023-04-11 15:26 ` Jeffrey Hugo
2023-04-11 15:29 ` Jeffrey Hugo
2023-04-11 16:31 ` Daniel Vetter
2023-04-11 17:18 ` Jeffrey Hugo
2023-04-11 18:21 ` Daniel Vetter
2023-04-11 18:37 ` Jeffrey Hugo [this message]
2023-04-11 18:47 ` Daniel Vetter
2023-04-12 6:11 ` Greg KH
-- strict thread matches above, loose matches on Subject: below --
2023-08-15 7:24 Stephen Rothwell
2023-08-15 15:21 ` Greg KH
2023-08-16 7:39 ` Maxime Ripard
2023-03-27 4:46 Stephen Rothwell
2023-03-27 7:33 ` Greg KH
2023-03-27 8:22 ` Stephen Rothwell
2023-03-27 9:15 ` Greg KH
2023-03-27 15:09 ` Vasily Gorbik
2023-01-30 4:28 Stephen Rothwell
2023-01-30 16:01 ` Greg KH
2023-01-30 16:23 ` Geert Uytterhoeven
2022-12-01 2:18 Stephen Rothwell
2022-11-30 1:38 Stephen Rothwell
2022-11-30 12:01 ` Greg KH
2022-11-28 2:36 Stephen Rothwell
2022-11-28 11:50 ` Greg KH
2022-11-28 12:44 ` Stephen Rothwell
2022-11-28 16:22 ` Greg KH
2022-11-28 19:46 ` Stephen Rothwell
2022-11-29 8:28 ` Greg KH
2022-11-28 17:35 ` Greg KH
2022-11-14 4:12 Stephen Rothwell
2022-11-14 8:46 ` Andy Shevchenko
2022-11-14 8:59 ` Greg KH
2021-07-23 5:09 Stephen Rothwell
2021-07-23 5:34 ` Uwe Kleine-König
2021-08-10 12:13 ` Geert Uytterhoeven
2021-08-12 20:51 ` Doug Anderson
2021-02-10 10:47 Stephen Rothwell
2021-02-10 18:18 ` Greg KH
2021-02-10 19:06 ` Saravana Kannan
2021-02-10 19:17 ` Saravana Kannan
2021-02-10 19:36 ` Greg KH
2021-02-10 20:14 ` Rob Herring
2021-02-10 20:43 ` Saravana Kannan
2021-02-10 20:59 ` Rob Herring
2020-09-09 5:47 Stephen Rothwell
2020-09-09 7:21 ` Greg KH
2020-09-09 21:47 ` Kees Cook
2020-09-09 7:28 ` Greg KH
2020-07-27 6:55 Stephen Rothwell
2020-07-27 10:17 ` Greg KH
2020-07-28 19:53 ` Kees Cook
2019-09-18 17:09 Mark Brown
2019-09-18 18:52 ` Greg KH
2019-09-18 19:03 ` Linus Torvalds
2019-09-18 19:07 ` Greg KH
2019-09-18 20:56 ` Suzuki K Poulose
2019-09-18 21:55 ` Mark Brown
2019-08-29 7:35 Stephen Rothwell
2019-08-29 10:10 ` Greg KH
2019-02-01 2:41 Stephen Rothwell
2019-02-01 14:18 ` Greg KH
2017-06-13 6:04 Stephen Rothwell
2017-06-13 6:20 ` Greg KH
2017-06-13 6:25 ` Stephen Rothwell
2017-06-13 6:30 ` Sergey Senozhatsky
2016-05-02 5:42 Stephen Rothwell
2016-05-02 11:40 ` William Breathitt Gray
2016-05-02 14:47 ` Greg KH
2015-05-21 7:52 Stephen Rothwell
2014-12-01 7:49 Stephen Rothwell
2014-05-28 8:03 Stephen Rothwell
2014-05-28 9:30 ` Jean Delvare
2014-03-12 0:51 Mark Brown
2014-03-12 1:50 ` Greg KH
2014-03-12 3:55 ` Benjamin Herrenschmidt
2014-03-12 11:37 ` Mark Brown
2014-03-12 19:59 ` Benjamin Herrenschmidt
2014-03-12 20:02 ` Tejun Heo
2014-03-12 20:14 ` Benjamin Herrenschmidt
2014-03-12 20:21 ` Tejun Heo
2014-03-13 0:37 ` Benjamin Herrenschmidt
2014-03-13 22:14 ` Benjamin Herrenschmidt
2014-03-14 13:00 ` Tejun Heo
2014-03-15 0:03 ` Greg KH
2014-03-15 2:57 ` Benjamin Herrenschmidt
2014-03-15 5:29 ` Greg KH
2014-03-15 7:14 ` Benjamin Herrenschmidt
2014-03-16 23:16 ` Stephen Rothwell
2014-03-17 18:36 ` Greg KH
2014-03-16 23:16 ` Stephen Rothwell
2014-03-17 18:33 ` Greg KH
2014-03-17 20:33 ` Benjamin Herrenschmidt
2014-03-17 21:56 ` Greg KH
2014-03-17 22:05 ` Tejun Heo
2014-03-17 22:21 ` Tejun Heo
2014-03-18 0:07 ` Stewart Smith
2014-03-18 0:22 ` Benjamin Herrenschmidt
2014-03-18 15:58 ` Tejun Heo
2014-03-18 0:00 ` Stewart Smith
2014-03-18 0:16 ` Benjamin Herrenschmidt
2014-03-17 8:28 ` Stephen Rothwell
2013-08-22 5:39 Stephen Rothwell
2013-08-22 15:36 ` Greg KH
2013-01-18 2:29 Stephen Rothwell
2013-01-18 4:13 ` Greg KH
2013-01-18 14:14 ` Kondratiev, Vladimir
2012-08-17 4:01 Stephen Rothwell
2012-08-17 13:25 ` Greg KH
2012-07-10 5:32 Stephen Rothwell
2012-07-10 9:03 ` Mark Brown
2012-07-10 11:14 ` Stephen Rothwell
2012-07-10 21:10 ` Linus Walleij
2012-05-01 4:59 Stephen Rothwell
2012-05-01 7:05 ` Bart Van Assche
2012-05-01 13:45 ` Greg KH
2012-04-19 4:48 Stephen Rothwell
2012-04-19 20:07 ` Greg KH
2012-04-20 2:19 ` Greg KH
2012-01-27 2:59 Stephen Rothwell
2012-01-27 15:24 ` Alan Stern
2012-01-27 22:36 ` Greg KH
2012-01-05 6:28 Stephen Rothwell
2012-01-05 23:20 ` Greg KH
2011-12-28 6:45 Stephen Rothwell
2012-01-03 16:21 ` Greg KH
2012-01-04 0:07 ` Kay Sievers
2012-01-04 0:31 ` Greg KH
2012-01-04 22:03 ` Kay Sievers
2012-01-04 23:13 ` Greg KH
2012-01-04 23:07 ` Greg KH
2012-01-04 23:48 ` Stephen Rothwell
2012-01-05 0:01 ` Kay Sievers
2012-01-05 0:17 ` Stephen Rothwell
2012-01-05 0:57 ` Josh Triplett
2011-12-15 6:37 Stephen Rothwell
2011-12-15 7:21 ` Greg KH
2011-12-15 10:10 ` Kay Sievers
2011-12-15 19:09 ` Greg KH
2011-12-14 4:48 Stephen Rothwell
2011-12-14 9:24 ` Peter Zijlstra
2011-12-14 15:03 ` Greg KH
2011-12-14 15:13 ` Peter Zijlstra
2011-12-14 17:01 ` Greg KH
2011-12-14 19:24 ` Greg KH
2011-11-21 1:57 Stephen Rothwell
2010-08-04 4:25 Stephen Rothwell
2010-08-04 4:44 ` Guenter Roeck
2010-08-04 15:52 ` Greg KH
2010-08-04 21:24 ` Jesse Barnes
2010-05-18 6:45 Stephen Rothwell
2010-05-18 7:00 ` Eric W. Biederman
2010-05-18 7:48 ` Al Viro
2010-05-18 7:54 ` Stephen Rothwell
2010-05-18 10:35 ` Eric W. Biederman
2010-05-18 15:02 ` Al Viro
2010-05-18 19:58 ` Eric W. Biederman
2010-05-18 21:25 ` Greg KH
2010-05-18 21:59 ` Eric W. Biederman
2010-05-18 6:44 Stephen Rothwell
2010-05-18 14:02 ` Greg KH
2010-05-18 14:04 ` Chris Wright
2010-05-18 21:06 ` Chris Wright
2010-05-18 21:24 ` Greg KH
2010-03-17 4:41 Stephen Rothwell
2010-03-17 7:21 ` Neil Brown
2010-03-17 15:51 ` Sage Weil
2010-03-17 22:30 ` Stephen Rothwell
2010-03-24 1:37 ` Neil Brown
2010-03-24 14:54 ` Sage Weil
2010-03-03 6:14 Stephen Rothwell
2010-03-03 6:31 ` Stephen Rothwell
2010-02-16 6:39 Stephen Rothwell
2010-02-16 9:01 ` Jean Delvare
2010-02-17 4:56 ` Greg KH
2010-02-18 4:27 ` Greg KH
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=086e08a2-13b3-870c-4b17-1fdc9d56d551@quicinc.com \
--to=quic_jhugo@quicinc.com \
--cc=airlied@redhat.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=greg@kroah.com \
--cc=jacek.lawrynowicz@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-next@vger.kernel.org \
--cc=mani@kernel.org \
--cc=ogabbay@kernel.org \
--cc=quic_pkanojiy@quicinc.com \
--cc=sfr@canb.auug.org.au \
/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 NNTP newsgroup(s).