Re: [PATCH] driver: svc-i3c-master: Kernel panic issue fixed

[Frank Li]

On Fri, Mar 07, 2025 at 12:34:53PM +0000, Manjunatha Venkatesh wrote:
>
>
> > -----Original Message-----
> > From: Frank Li <frank.li@xxxxxxx>
> > Sent: Thursday, March 6, 2025 9:01 PM
> > To: Manjunatha Venkatesh <manjunatha.venkatesh@xxxxxxx>
> > Cc: miquel.raynal@xxxxxxxxxxx; conor.culhane@xxxxxxxxxxx;
> > alexandre.belloni@xxxxxxxxxxx; linux-i3c@xxxxxxxxxxxxxxxxxxx; linux-
> > kernel@xxxxxxxxxxxxxxx; rvmanjumce@xxxxxxxxx
> > Subject: Re: [PATCH] driver: svc-i3c-master: Kernel panic issue fixed
> >
> > On Thu, Mar 06, 2025 at 08:31:29AM +0000, Manjunatha Venkatesh wrote:
> > > Hi Frank,
> > >
> > > Please find my inline comments.
> > >
> > > Regards,
> > > Manju
> >
> > Avoid top post for linux kernel mail list.
> >
> > >
> > > > -----Original Message-----
> > > > From: Frank Li <frank.li@xxxxxxx>
> > > > Sent: Friday, February 28, 2025 9:19 PM
> > > > To: Manjunatha Venkatesh <manjunatha.venkatesh@xxxxxxx>
> > > > Cc: miquel.raynal@xxxxxxxxxxx; conor.culhane@xxxxxxxxxxx;
> > > > alexandre.belloni@xxxxxxxxxxx; linux-i3c@xxxxxxxxxxxxxxxxxxx; linux-
> > > > kernel@xxxxxxxxxxxxxxx; rvmanjumce@xxxxxxxxx
> > > > Subject: Re: [PATCH] driver: svc-i3c-master: Kernel panic issue
> > > > fixed
> > > >
> > > > Subject should be "Fix read from unreadable memory at
> > > > svc_i3c_master_ibi_work()"
> > > > Or like below pattern:
> > > > 	Do ... to fix ...
> > > >
> > > [Manjunatha Venkatesh] Tittle updated as per the review comment.
> > > > On Mon, Feb 24, 2025 at 08:47:54AM +0530, Manjunatha Venkatesh
> > wrote:
> > > > > I3C Master driver svc-i3c-master
> > > >
> > > > Needn't this line.
> > > >
> > > [Manjunatha Venkatesh] Removed this line and submitted the new patch
> > > > >
> > > > > As part of I3C driver probing sequence for particular device
> > > > > instance, While adding to queue it is trying to access ibi
> > > > > variable of dev which is not yet initialized causing "Unable to
> > > > > handle kernel read from unreadable memory" resulting in kernel panic.
> > > >
> > > > Can you provide more detail information why/how this happen? which
> > > > varible read cause this problem?
> > > >
> > > [Manjunatha Venkatesh] : Below is the sequence where this issue
> > happened.
> > > 1. During boot up sequence IBI is received at host  from the slave device
> > before requesting for IBI, Usually will request IBI by calling
> > i3c_device_request_ibi() during probe of slave driver.
> > > 2. Since master code trying to access IBI Variable for the particular device
> > instance before actually it initialized by slave driver, due to this randomly
> > accessing the address and causing kernel panic.
> > > 3. i3c_device_request_ibi() function invoked by the slave driver where dev-
> > >ibi = ibi; assigned as part of function call i3c_dev_request_ibi_locked().
> > > 4. But when IBI request sent by slave device, master code  trying to access
> > this variable before its initialized due to this race condition situation kernel
> > panic happened.
> >
> > Write above reproduce steps to commit message.
> >
> [Manjunatha Venkatesh] : Next commit will update this description.
> > I think it is common problem for i3c framework.
> >
> > You'd better change i3c_master_queue_ibi()
> >
> > void i3c_master_queue_ibi(struct i3c_dev_desc *dev, struct i3c_ibi_slot *slot)
> > {
> > 	if (!slot)
> > 		return;
> >
> >         atomic_inc(&dev->ibi->pending_ibis);
> >         queue_work(dev->ibi->wq, &slot->work); }
> >
> [Manjunatha Venkatesh] : Actual issue we have faced with respect to 'dev->ibi' variable.

How about
if (!dev->ibi || !slot) {
	dev_warning("...");
	return;
}

Frank

> Pointer variable 'slot' is with respect IBI received from slave device context and 'dev->ibi' is with respect to initialization of slave driver these 2 are asynchronous events.
> dev->ibi = ibi; is initialized at i3c_dev_request_ibi_locked(), this is invoked as part of slave driver IBI init sequence, this is what we want to check before processing the IBI request from the Slave device.
> master->ibi.slots[i] this is where all device ibi instance list are saved, so here our slave driver init IBI stored with respect dev->ibi index.
> So adding 'if (!slot)' will not help to resolve this issue.


Please wrap around at 80 char to read easily.


>
> > Frank
> > >
> > >
> > > > Need fix tags and cc stable.
> > > >
> > > [Manjunatha Venkatesh] : fix tags and cc stable updated in the commit
> > message.
> > >
> [Manjunatha Venkatesh] : Last commit message missed to update fix tags and cc stable.
> Will update that information as part of next commit message.
> > > > Frank
> > > > >
> > > > > Signed-off-by: Manjunatha Venkatesh
> > <manjunatha.venkatesh@xxxxxxx>
> > > > > ---
> > > > >  drivers/i3c/master/svc-i3c-master.c | 7 +++++--
> > > > >  1 file changed, 5 insertions(+), 2 deletions(-)
> > > > >
> > > > > diff --git a/drivers/i3c/master/svc-i3c-master.c
> > > > > b/drivers/i3c/master/svc-i3c-master.c
> > > > > index d6057d8c7dec..98c4d2e5cd8d 100644
> > > > > --- a/drivers/i3c/master/svc-i3c-master.c
> > > > > +++ b/drivers/i3c/master/svc-i3c-master.c
> > > > > @@ -534,8 +534,11 @@ static void svc_i3c_master_ibi_work(struct
> > > > work_struct *work)
> > > > >  	switch (ibitype) {
> > > > >  	case SVC_I3C_MSTATUS_IBITYPE_IBI:
> > > > >  		if (dev) {
> > > > > -			i3c_master_queue_ibi(dev, master->ibi.tbq_slot);
> > > > > -			master->ibi.tbq_slot = NULL;
> > > > > +			data = i3c_dev_get_master_data(dev);
> > > > > +			if (master->ibi.slots[data->ibi]) {
> > > > > +				i3c_master_queue_ibi(dev, master-
> > > > >ibi.tbq_slot);
> > > > > +				master->ibi.tbq_slot = NULL;
> > > > > +			}
> > > > >  		}
> > > > >  		svc_i3c_master_emit_stop(master);
> > > > >  		break;
> > > > > --
> > > > > 2.46.1
> > > > >
> > > > >
> > > > > --
> > > > > linux-i3c mailing list
> > > > > linux-i3c@xxxxxxxxxxxxxxxxxxx
> > > > > http://lists.infradead.org/mailman/listinfo/linux-i3c