Re: PROBLEM: Firmware loader fallback mechanism no longer works with sendfile

[Siddharth Gupta]

On 1/15/2021 8:20 AM, Greg KH wrote:
> On Tue, Jan 12, 2021 at 10:31:26AM -0800, Siddharth Gupta wrote:
> > On 1/8/2021 6:44 AM, Greg KH wrote:
> > > On Thu, Jan 07, 2021 at 02:03:47PM -0800, Siddharth Gupta wrote:
> > > > On 1/6/2021 2:33 AM, Greg KH wrote:
> > > > > > > > Since the binary attributes don't support splice_{read,write} functions the
> > > > > > > > calls to splice_{read,write} used the default kernel_{read,write} functions.
> > > > > > > > With the above change this results in an -EINVAL return from
> > > > > > > > do_splice_from[4].
> > > > > > > >
> > > > > > > > This essentially means that sendfile will not work for any binary attribute
> > > > > > > > in the sysfs.
> > > > > > > Have you tried fixing this with a patch much like what we did for the
> > > > > > > proc files that needed this?  If not, can you?
> > > > > > I am not aware of this fix, could you provide me a link for reference? I
> > > > > > will try it out.
> > > > > Look at the series of commits starting at fe33850ff798 ("proc: wire up
> > > > > generic_file_splice_read for iter ops") for how this was fixed in procfs
> > > > > as an example of what also needs to be done for binary sysfs files.
> > > > I tried to follow these fixes, but I am unfamiliar with fs code. I don't see
> > > > the generic_file_splice_write function anymore on newer kernels, also AFAICT
> > > > kernfs_ops does not define {read,write}_iter operations. If the solution is
> > > > simple and someone could provide the patches I would be happy to test them
> > > > out. If not, some more information about how to proceed would be nice.
> > > Can you try this tiny patch out below?
> > Sorry for the delay, I tried out the patch, but I am still seeing the error.
> > Please take a look at these logs with
> > android running in the userspace[1]:
> >
> > [   62.295056][  T249] remoteproc remoteproc1: powering up
> > xxxxxxxx.remoteproc-cdsp
> > [   62.304138][  T249] remoteproc remoteproc1: Direct firmware load for
> > cdsp.mdt failed with error -2
> > [   62.312976][  T249] remoteproc remoteproc1: Falling back to sysfs
> > fallback for: cdsp.mdt
> > [   62.469748][  T394] ueventd: firmware: loading 'cdsp.mdt' for '/devices/platform/soc/xxxxxxxx.remoteproc-cdsp/remoteproc/remoteproc1/cdsp.mdt'
> > [   62.498700][  T394] ueventd: firmware: sendfile failed { '/sys/devices/platform/soc/xxxxxxxx.remoteproc-cdsp/remoteproc/remoteproc1/cdsp.mdt',
> > 'cdsp.mdt' }: Invalid argument
> >
> > Thanks,
> > Sid
> >
> > [1]: https://android.googlesource.com/platform/system/core/+/refs/heads/master/init/firmware_handler.cpp#57
> Thanks for letting me know.  I'll try to work on this on Monday and add
> splice to the in-kernel firmware testing suite, as it would have caught
> this...
If my understanding is correct these default (or generic if you prefer)
functions require the write_iter[1] and read_iter[2] file ops to be 
defined for
kernfs_file_fops. Are there any similar default {write,read}_iter 
functions that
I can use, or do these need to be thin wrappers for the read function[3] 
for bin
attributes itself?

Thanks,
Sid

[1]: 
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/splice.c#n686
[2]: 
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/splice.c#n311
[3]: 
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/sysfs/file.c#n235
> greg k-h