linux-kernel - Re: [PATCH] remoteproc: elf_loader: Update resource table name check

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Mon, 12 Dec 2022 14:04:44 -0500
From:   Stephen Boyd <swboyd@...omium.org>
To:     Srinivasa Rao Mandadapu <quic_srivasam@...cinc.com>,
        agross@...nel.org, andersson@...nel.org, bgoswami@...cinc.com,
        broonie@...nel.org, devicetree@...r.kernel.org,
        judyhsiao@...omium.org, krzysztof.kozlowski@...aro.org,
        lgirdwood@...il.com, linux-arm-msm@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-remoteproc@...r.kernel.org,
        mathieu.poirier@...aro.org, perex@...ex.cz, quic_plai@...cinc.com,
        quic_rohkumar@...cinc.com, robh+dt@...nel.org,
        srinivas.kandagatla@...aro.org, tiwai@...e.com
Subject: Re: [PATCH] remoteproc: elf_loader: Update resource table name check

Quoting Srinivasa Rao Mandadapu (2022-12-12 05:49:29)
>
> On 12/10/2022 2:22 AM, Stephen Boyd wrote:
> Thanks for your time Stephen!!!
> > Quoting Srinivasa Rao Mandadapu (2022-12-08 05:40:54)
> >> On 12/7/2022 7:25 AM, Stephen Boyd wrote:
> >> Thanks for Your Time Stephen!!!
> >>> Quoting Srinivasa Rao Mandadapu (2022-12-01 04:20:48)
> >>>> Update resource table name check with sub string search instead of
> >>>> complete string search.
> >>>> In general Qualcomm binary contains, section header name
> >>>> (e.g. .resource_table), amended with extra string to differentiate
> >>>> with other sections.
> >>>> So far Android adsp binaries are being authenticated using TZ,
> >>>> hence this mismatch hasn't created any problem.
> >>>> In recent developments, ADSP binary is being used in Chrome based
> >>>> platforms, which doesn't have TZ path, hence resource table is
> >>>> required for memory sandboxing.
> >>>>
> >>> Does this need a Fixes tag?
> >> I don't think so. I feel It's kind of enhancement.
> >>>> Signed-off-by: Srinivasa Rao Mandadapu <quic_srivasam@...cinc.com>
> >>>> ---
> >>>>    drivers/remoteproc/remoteproc_elf_loader.c | 2 +-
> >>>>    1 file changed, 1 insertion(+), 1 deletion(-)
> >>>>
> >>>> diff --git a/drivers/remoteproc/remoteproc_elf_loader.c b/drivers/remoteproc/remoteproc_elf_loader.c
> >>>> index 5a412d7..0feb120 100644
> >>>> --- a/drivers/remoteproc/remoteproc_elf_loader.c
> >>>> +++ b/drivers/remoteproc/remoteproc_elf_loader.c
> >>>> @@ -272,7 +272,7 @@ find_table(struct device *dev, const struct firmware *fw)
> >>>>                   u64 offset = elf_shdr_get_sh_offset(class, shdr);
> >>>>                   u32 name = elf_shdr_get_sh_name(class, shdr);
> >>>>
> >>>> -               if (strcmp(name_table + name, ".resource_table"))
> >>>> +               if (!strstr(name_table + name, ".resource_table"))
> >>> Was the strcmp not working before because the 'name_table' has something
> >>> else in it? It really looks like your elf file is malformed.
> >> Actually, DSP binary is prepared by combining different elfs. So Section
> >> header names are appended with
> >>
> >> something else to distinguish same section name of different elfs, by
> >> using some Qualcomm specific QURT scripts.
> >> Hence final binary contains resource_table name appended with module
> >> specific name.
> >>
> >> So this patch is required to handle such modified name.
> >>
> > Can you clarify how the section header name looks? Probably you can
> > objdump the section here and provide that information to help us
> > understand.
>
> Here is the Section header info.
>
> $ readelf -SW bootimage_relocflag_kodiak.adsp.prodQ.pbn
> There are 65 section headers, starting at offset 0x434:
> readelf: Error: File contains multiple dynamic symbol tables
>
[...]
>    [60] .start.ac_bin_process PROGBITS
>    [61] .resource_table.ac_bin_process PROGBITS

Cool, the readelf output is helpful. Please rewrite the commit text to
include this detail. It wasn't obvious to me what 'amended' meant. You
probably mean "appended", which clarifies that it has a string added to
the end. I'm also not sure why TZ or not TZ matters for the resource
table section. It may be meaningful to you, but to others it doesn't
have any relation to this resource table appending scheme so it is not
helpful by itself.

Either way, this is not up to me as I'm not the maintainer of
remoteproc. I peeked at the documentation, but this section
name isn't clearly defined. It seems to just be how it has been for a
long time. Maybe you can also update the documentation
(Documentation/staging/remoteproc.rst) to indicate that this elf section
can have anything appended after it, but it must start with
".resource_table"? That would help everyone. And I don't know why that's
in the staging directory. Bjorn?

Finally, I'd prefer the use of strstarts() instead so it is clear what
you're trying to implement.