Re: [PATCH v2] staging: vme_user: fix flush_image leak in tsi148 bridge

[Nguyễn Công]

On Thu, Aug 13, 2026 at 7:26 PM Greg Kroah-Hartman
<gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On Thu, Aug 13, 2026 at 07:10:57PM +0700, Cong Nguyen wrote:
> > When error checking is enabled (err_chk=1), tsi148_probe() allocates an
> > extra master window resource, tsi148_device->flush_image, which is used
> > to flush posted writes by reading back over the VME bus. Unlike the
> > regular master windows, this resource is not linked into
> > tsi148_bridge->master_resources.
> >
> > Because it is not on any resource list, it is freed neither by the probe
> > error path (which only walks the master_resources list) nor by
> > tsi148_remove(), so it is leaked whenever err_chk is set and either
> > probe fails after the allocation or the device is unbound / the module
> > is unloaded.
> >
> > Free flush_image in both the probe error path and tsi148_remove(). This
> > is safe when err_chk is disabled: tsi148_device is allocated with
> > kzalloc() so flush_image is NULL and kfree(NULL) is a no-op.
> >
> > Fixes: d22b8ed9a3b0 ("Staging: vme: add Tundra TSI148 VME-PCI Bridge driver")
> > Cc: stable@xxxxxxxxxxxxxxx
> > Assisted-by: Claude:claude-opus-4
> > Signed-off-by: Cong Nguyen <congnt264@xxxxxxxxx>
> > ---
> > Changes in v2:
> > - Add Assisted-by: tag (Greg Kroah-Hartman). No code changes.
>
> You have read this:
>         https://lore.kernel.org/all/2026080354-skater-urgent-31b2@gregkh/T/#u
> right?
>
> If not, please do so.
>
> thanks,
>
> greg k-h

I have now, thanks. I don't have TSI148 hardware to test on, so this
change doesn't meet the bar you describe for LLM-assisted staging
patches. Please drop it.

I'll hold off on further drivers/staging/ submissions unless I can test
them on the actual hardware. Sorry for the noise, and thanks for the
clear guidance.

Cong