Re: linuxnext-2019127 edac warns (was Re: edac KASAN warning in experimental arm64 allmodconfig boot)

[Robert Richter]

On 02.12.19 10:23:29, John Garry wrote:
> On 28/11/2019 21:12, Robert Richter wrote:
> > On 27.11.19 17:07:33, John Garry wrote:
> > > [   22.104498] BUG: KASAN: use-after-free in
> > > edac_remove_sysfs_mci_device+0x148/0x180
> > 
> > It is triggered in edac_remove_sysfs_mci_device().
> > 
> > device_unregister(&dimm->dev) not only removes the sysfs entry, it
> > also frees the dimm struct in dimm_attr_release(). When incrementing
> > the loop in mci_for_each_dimm(), the dimm struct is accessed again
> > which causes the use-after-free. But, the dimm struct shouln'd be
> > released here already.
> > 
> > edac_remove_sysfs_mci_device() should not release the devices at this
> > point. We need clean release functions for mci and dimm_info and
> > refcounts to protect pdev/dev mappings. And mci_for_each_dimm() must
> > be checked how it handles device removals and if it is safe.
> > 
> > Let's see how this can be fixed.
> > 
> > Thanks for reporting the issue.
> 
> Fine, and would any fix also deal with the v5.4 mem leak which I mentioned
> also?

Yes, I have identified the leaks:

# grep edac /sys/kernel/debug/kmemleak | sort | uniq -c
      1     [<000000003c0f58f9>] edac_mc_alloc+0x3bc/0x9d0	# mci->csrows
     16     [<00000000bb932dc0>] edac_mc_alloc+0x49c/0x9d0	# csr->channels
     16     [<00000000e2734dba>] edac_mc_alloc+0x518/0x9d0	# csr->channels[chn]
      1     [<00000000eb040168>] edac_mc_alloc+0x5c8/0x9d0	# mci->dimms
     34     [<00000000ef737c29>] ghes_edac_register+0x1c8/0x3f8	# see edac_mc_alloc()

Thanks,

-Robert