Re: [PATCH v2] pps: generators: tio: fix platform_set_drvdata()

[Andy Shevchenko]

On Mon, Mar 17, 2025 at 09:24:52PM +0530, Raag Jadav wrote:
> Currently driver_data user is expecting a pointer to pps_tio while

struct pps_tio

> platform_set_drvdata() is setting a double pointer to it.

"...which will point to the local stack of the probe function."

> Make them
> consistent and fix illegal memory access in driver ->remove() path.
> 
> [  156.254066] BUG: unable to handle page fault for address: ffffc9000117b738

> [  156.254099] #PF: supervisor read access in kernel mode
> [  156.254111] #PF: error_code(0x0000) - not-present page

These lines are not needed...

> [  156.254197] RIP: 0010:hrtimer_active+0x2b/0x60
> [  156.254367] Call Trace:

> [  156.254375]  <TASK>
> [  156.254382]  ? show_regs+0x6d/0x80
> [  156.254393]  ? __die+0x29/0x70
> [  156.254402]  ? page_fault_oops+0x15f/0x4e0
> [  156.254415]  ? hrtimer_active+0x2b/0x60
> [  156.254425]  ? search_exception_tables+0x65/0x70
> [  156.254437]  ? kernelmode_fixup_or_oops.constprop.0+0x61/0x80
> [  156.254451]  ? __bad_area_nosemaphore+0x195/0x2c0
> [  156.254462]  ? __lock_acquire+0xaaf/0x2840
> [  156.254475]  ? bad_area_nosemaphore+0x16/0x20
> [  156.254486]  ? do_kern_addr_fault.part.0+0x64/0x80
> [  156.254498]  ? exc_page_fault+0x190/0x2c0
> [  156.254511]  ? asm_exc_page_fault+0x2b/0x30
> [  156.254527]  ? __pfx_pps_gen_tio_remove+0x10/0x10 [pps_gen_tio]

...neither these.

> [  156.254541]  ? hrtimer_active+0x2b/0x60
> [  156.254551]  hrtimer_cancel+0x19/0x50
> [  156.254561]  pps_gen_tio_remove+0x1e/0x80 [pps_gen_tio]

As I said, ~3-5, okay maybe 7 lines is enough.

> Fixes: c89755d1111f ("pps: generators: Add PPS Generator TIO Driver")
> Signed-off-by: Raag Jadav <raag.jadav@xxxxxxxxx>
> Acked-by: Rodolfo Giometti <giometti@xxxxxxxxxxxx>

-- 
With Best Regards,
Andy Shevchenko