Re: CVE-2024-26827: i2c: qcom-geni: Correct I2C TRE sequence

From: Greg Kroah-Hartman
Date: Thu Apr 18 2024 - 10:52:30 EST


On Thu, Apr 18, 2024 at 04:44:33PM +0200, Jean Delvare wrote:
> On Thu, 2024-04-18 at 15:05 +0200, Greg Kroah-Hartman wrote:
> > On Thu, Apr 18, 2024 at 02:56:33PM +0200, Jean Delvare wrote:
> > > Hi Greg,
> > >
> > > On Wed, 2024-04-17 at 11:44 +0200, Greg Kroah-Hartman wrote:
> > > > Description
> > > > ===========
> > > >
> > > > In the Linux kernel, the following vulnerability has been resolved:
> > > >
> > > > i2c: qcom-geni: Correct I2C TRE sequence
> > > >
> > > > For i2c read operation in GSI mode, we are getting timeout
> > > > due to malformed TRE basically incorrect TRE sequence
> > > > in gpi(drivers/dma/qcom/gpi.c) driver.
> > > > (...)
> > >
> > > I was assigned the task to backport this security fix to the SUSE
> > > kernels. However, from the description, I fail to see how this fix
> > > qualifies as a security fix. I can't find the reason why a CVE was
> > > assigned to the issue.
> > >
> > > What is the considered attack vector? Or if there is no attack vector,
> > > what consequence does this bug have, which would put the system
> > > security at stake?
> >
> > We reviewed this commit as fitting the fact that timeouts due to
> > malformed messages would fit into the definition of "vulnerability" in
> > the CVE world as it would cause a system to incure "negative impact to
> > confidentiality, integrity, or availability".
>
> If the timeout could be triggered on purpose, then I would agree, as
> this could possibly be used for a denial-of-service type of attack. But
> this isn't the case here.
>
> All we have is a failure to read data from a random I2C device due to
> an incorrect programming of the I2C controller. Simple lack of
> functionality.

Thanks for the explaination and looking into it.

> > If as the i2c maintainer, you don't think this would be the case, we
> > will be glad to revoke this CVE and just mark it down as a "normal
> > bugfix".
>
> Yes, please.

Now rejected, thanks for letting us know.

greg k-h