Re: WARNING in bpf_bprintf_prepare

[Jiri Olsa <olsajiri@gmail.com>]

On Tue, Nov 15, 2022 at 09:48:27AM +0800, Hao Sun wrote:

SNIP

> > > > Hi Jirka,
> > > >
> > > > I've tested the proposed patch, the warning from bpf_bprintf_prepare will not
> > > > be triggered with the patch, but the reproducer can still trigger the following
> > > > warning. My test was conducted on:
> > > >
> > > > commit:  f67dd6ce0723 Merge tag 'slab-for-6.1-rc4-fixes'
> > > > git tree:   upstream
> > > > kernel config: https://pastebin.com/raw/sE5QK5HL
> > > > C reproducer: https://pastebin.com/raw/X96ASi27
> > > > console log *before* the patch: https://pastebin.com/raw/eSCUNFrd
> > > > console log *after* the patch: https://pastebin.com/raw/tzcmdWZt
> > >
> > > thanks for testing.. I can't reproduce this for some reason
> > >
> > > I'll check some more and perhaps go with denying bpf attachment
> > > for this tracepoint as Alexei suggeste
> >
> > the change below won't allow to attach bpf program with any printk
> > helper in contention_begin and bpf_trace_printk tracepoints
> >
> > I still need to test it on the machine that reproduced the issue
> > for me.. meanwhile any feedback is appreciated
> >
> 
> Hi,
> 
> Tested on my machine, the C reproducer won't trigger any issue
> this time with the patch. The test was conducted on:
> 
> commit:  f67dd6ce0723 Merge tag 'slab-for-6.1-rc4-fixes'
> git tree:   upstream
> kernel config: https://pastebin.com/raw/sE5QK5HL
> C reproducer: https://pastebin.com/raw/X96ASi27
> full console log *before* the patch: https://pastebin.com/raw/n3x55RDr
> full console log *after* the patch: https://pastebin.com/raw/7HdxnCnL

thanks, looks good on my end as well

jirka