Re: [PATCH] scsi: ufs: core: move some irq handling back to hardirq (with time limit)

[André Draszik]

On Thu, 2025-07-24 at 13:54 +0200, Neil Armstrong wrote:
> On 24/07/2025 13:44, André Draszik wrote:
> > On Thu, 2025-07-24 at 10:54 +0100, André Draszik wrote:
> > > fio results on Pixel 6:
> > >    read / 1 job     original    after    this commit
> > >      min IOPS        4,653.60   2,704.40    3,902.80
> > >      max IOPS        6,151.80   4,847.60    6,103.40
> > >      avg IOPS        5,488.82   4,226.61    5,314.89
> > >      cpu % usr           1.85       1.72        1.97
> > >      cpu % sys          32.46      28.88       33.29
> > >      bw MB/s            21.46      16.50       20.76
> > > 
> > >    read / 8 jobs    original    after    this commit
> > >      min IOPS       18,207.80  11,323.00   17,911.80
> > >      max IOPS       25,535.80  14,477.40   24,373.60
> > >      avg IOPS       22,529.93  13,325.59   21,868.85
> > >      cpu % usr           1.70       1.41        1.67
> > >      cpu % sys          27.89      21.85       27.23
> > >      bw MB/s            88.10      52.10       84.48
> > > 
> > >    write / 1 job    original    after    this commit
> > >      min IOPS        6,524.20   3,136.00    5,988.40
> > >      max IOPS        7,303.60   5,144.40    7,232.40
> > >      avg IOPS        7,169.80   4,608.29    7,014.66
> > >      cpu % usr           2.29       2.34        2.23
> > >      cpu % sys          41.91      39.34       42.48
> > >      bw MB/s            28.02      18.00       27.42
> > > 
> > >    write / 8 jobs   original    after    this commit
> > >      min IOPS       12,685.40  13,783.00   12,622.40
> > >      max IOPS       30,814.20  22,122.00   29,636.00
> > >      avg IOPS       21,539.04  18,552.63   21,134.65
> > >      cpu % usr           2.08       1.61        2.07
> > >      cpu % sys          30.86      23.88       30.64
> > >      bw MB/s            84.18      72.54       82.62
> > 
> > Given the severe performance drop introduced by the culprit
> > commit, it might make sense to instead just revert it for
> > 6.16 now, while this patch here can mature and be properly
> > reviewed. At least then 6.16 will not have any performance
> > regression of such a scale.
> 
> The original change was designed to stop the interrupt handler
> to starve the system and create display artifact and cause
> timeouts on system controller submission. While imperfect,
> it would require some fine tuning for smaller controllers
> like on the Pixel 6 that when less queues.

Well, the patch has solved one problem by creating another problem.
I don't think that's how things are normally done. A 40% bandwidth
and IOPS drop is not negligible.

And while I am referencing Pixel 6 above as it's the only device
I have available to test, I suspect all < v4 controllers / devices
are affected in a similar way, given the nature of the change.


Cheers,
Andre'