On Wed, Mar 10, 2021 at 09:21:55PM +0100, Greg Kroah-Hartman wrote:

On Wed, Mar 10, 2021 at 10:10:41PM +0200, Leon Romanovsky wrote:

quoted

On Wed, Mar 10, 2021 at 01:09:06PM -0600, Bjorn Helgaas wrote:

quoted

On Sun, Mar 07, 2021 at 10:55:24AM -0800, Alexander Duyck wrote:

quoted

On Sun, Feb 28, 2021 at 11:55 PM Leon Romanovsky [off-list ref] wrote:

quoted

From: Leon Romanovsky <leonro@nvidia.com>

@Alexander Duyck, please update me if I can add your ROB tag again
to the series, because you liked v6 more.

Thanks

---------------------------------------------------------------------------------
Changelog
v7:
 * Rebase on top v5.12-rc1
 * More english fixes
 * Returned to static sysfs creation model as was implemented in v0/v1.

<...>

quoted

  2) Should a VF sysfs file use the PF to implement this?

     Can you elaborate on your idea here?  I guess
     pci_iov_sysfs_link() makes a "virtfnX" link from the PF to the
     VF, and you're thinking we could also make a "virtfnX_msix_count"
     in the PF directory?  That's a really interesting idea.

I want to remind that we are talking about mlx5 devices that support
upto 255 VFs and they indeed are used to their limits. So seeing 255
links of virtfnX_msix_count in the same directory looks too much unpleasant
to me.

255 files are nothing, if that's what the hardware supports, what is the
problem?  If it's "unpleasant", go complain to the hardware designers :)

It is 255 same files that every SR-IOV user will see in /sys/bus/pci/devices/*/
folder, unless we will do dynamic creation of those files and this is something
that Bjorn didn't like in v7.

So instead of complaining to the hardware designers, I will complain here.
I probably implemented all possible variants already. :)

Thanks

greg k-h