I'm looking into a linkage failure for one of our device kernels, and
it seems that genksyms isn't producing a hash value correctly for
aggregate definitions that contain __attribute__s like
__attribute__((packed)).
Example:
$ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
Defn for struct foo == <struct foo { int bar ; } >
Hash table occupancy 1/4096 = 0.000244141
$ echo 'struct __attribute__((packed)) foo { int bar; };' |
./scripts/genksyms/genksyms -d
Hash table occupancy 0/4096 = 0
I assume the __attribute__ part isn't being parsed correctly (looks
like genksyms is a lex/yacc based C parser).
The issue we have in our out of tree driver (*sadface*) is basically a
EXPORT_SYMBOL'd function whose signature contains a packed struct.
Theoretically, there should be nothing wrong with exporting a function
that requires packed structs, and this is just a bug in the lex/yacc
based parser, right? I assume that not having CONFIG_MODVERSIONS
coverage of packed structs in particular could lead to potentially
not-fun bugs? Or is using packed structs in exported function symbols
with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
missed?
--
Thanks,
~Nick Desaulniers
On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
<[email protected]> wrote:
>
> I'm looking into a linkage failure for one of our device kernels, and
> it seems that genksyms isn't producing a hash value correctly for
> aggregate definitions that contain __attribute__s like
> __attribute__((packed)).
>
> Example:
> $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
> Defn for struct foo == <struct foo { int bar ; } >
> Hash table occupancy 1/4096 = 0.000244141
> $ echo 'struct __attribute__((packed)) foo { int bar; };' |
> ./scripts/genksyms/genksyms -d
> Hash table occupancy 0/4096 = 0
>
> I assume the __attribute__ part isn't being parsed correctly (looks
> like genksyms is a lex/yacc based C parser).
>
> The issue we have in our out of tree driver (*sadface*) is basically a
> EXPORT_SYMBOL'd function whose signature contains a packed struct.
>
> Theoretically, there should be nothing wrong with exporting a function
> that requires packed structs, and this is just a bug in the lex/yacc
> based parser, right? I assume that not having CONFIG_MODVERSIONS
> coverage of packed structs in particular could lead to potentially
> not-fun bugs? Or is using packed structs in exported function symbols
> with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
> missed?
Ah, looks like I'm late to the party:
https://lwn.net/Articles/707520/
--
Thanks,
~Nick Desaulniers
Nick Desaulniers's on August 27, 2019 8:57 am:
> On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
> <[email protected]> wrote:
>>
>> I'm looking into a linkage failure for one of our device kernels, and
>> it seems that genksyms isn't producing a hash value correctly for
>> aggregate definitions that contain __attribute__s like
>> __attribute__((packed)).
>>
>> Example:
>> $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
>> Defn for struct foo == <struct foo { int bar ; } >
>> Hash table occupancy 1/4096 = 0.000244141
>> $ echo 'struct __attribute__((packed)) foo { int bar; };' |
>> ./scripts/genksyms/genksyms -d
>> Hash table occupancy 0/4096 = 0
>>
>> I assume the __attribute__ part isn't being parsed correctly (looks
>> like genksyms is a lex/yacc based C parser).
>>
>> The issue we have in our out of tree driver (*sadface*) is basically a
>> EXPORT_SYMBOL'd function whose signature contains a packed struct.
>>
>> Theoretically, there should be nothing wrong with exporting a function
>> that requires packed structs, and this is just a bug in the lex/yacc
>> based parser, right? I assume that not having CONFIG_MODVERSIONS
>> coverage of packed structs in particular could lead to potentially
>> not-fun bugs? Or is using packed structs in exported function symbols
>> with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
>> missed?
>
> Ah, looks like I'm late to the party:
> https://lwn.net/Articles/707520/
Yeah, would be nice to do something about this.
IIRC (without re-reading it all), in theory distros would be okay
without modversions if they could just provide their own explicit
versioning. They take care about ABIs, so they can version things
carefully if they had to change.
I think we left that on hold because some of the bigger distros were
heading into releases and we didn't care to cause pain. I wonder if
we could try again.
What's your requirement for versioning?
Thanks,
Nick
Hi.
On Tue, Aug 27, 2019 at 6:59 PM Nicholas Piggin <[email protected]> wrote:
>
> Nick Desaulniers's on August 27, 2019 8:57 am:
> > On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
> > <[email protected]> wrote:
> >>
> >> I'm looking into a linkage failure for one of our device kernels, and
> >> it seems that genksyms isn't producing a hash value correctly for
> >> aggregate definitions that contain __attribute__s like
> >> __attribute__((packed)).
> >>
> >> Example:
> >> $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
> >> Defn for struct foo == <struct foo { int bar ; } >
> >> Hash table occupancy 1/4096 = 0.000244141
> >> $ echo 'struct __attribute__((packed)) foo { int bar; };' |
> >> ./scripts/genksyms/genksyms -d
> >> Hash table occupancy 0/4096 = 0
> >>
> >> I assume the __attribute__ part isn't being parsed correctly (looks
> >> like genksyms is a lex/yacc based C parser).
> >>
> >> The issue we have in our out of tree driver (*sadface*) is basically a
> >> EXPORT_SYMBOL'd function whose signature contains a packed struct.
> >>
> >> Theoretically, there should be nothing wrong with exporting a function
> >> that requires packed structs, and this is just a bug in the lex/yacc
> >> based parser, right? I assume that not having CONFIG_MODVERSIONS
> >> coverage of packed structs in particular could lead to potentially
> >> not-fun bugs? Or is using packed structs in exported function symbols
> >> with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
> >> missed?
> >
> > Ah, looks like I'm late to the party:
> > https://lwn.net/Articles/707520/
>
> Yeah, would be nice to do something about this.
modversions is ugly, so it would be great if we could dump it.
> IIRC (without re-reading it all), in theory distros would be okay
> without modversions if they could just provide their own explicit
> versioning. They take care about ABIs, so they can version things
> carefully if they had to change.
We have not provided any alternative solution for this, haven't we?
In your patch (https://lwn.net/Articles/707729/),
you proposed CONFIG_MODULE_ABI_EXPLICIT.
If it is good enough for distros, we merge it first,
give them time to migrate over to it, then finally remove modversions??
> I think we left that on hold because some of the bigger distros were
> heading into releases and we didn't care to cause pain. I wonder if
> we could try again.
I agree.
>
> What's your requirement for versioning?
I added Ben Hutchings to CC.
>
> Thanks,
> Nick
--
Best Regards
Masahiro Yamada
Masahiro Yamada's on August 27, 2019 8:49 pm:
> Hi.
>
> On Tue, Aug 27, 2019 at 6:59 PM Nicholas Piggin <[email protected]> wrote:
>>
>> Nick Desaulniers's on August 27, 2019 8:57 am:
>> > On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
>> > <[email protected]> wrote:
>> >>
>> >> I'm looking into a linkage failure for one of our device kernels, and
>> >> it seems that genksyms isn't producing a hash value correctly for
>> >> aggregate definitions that contain __attribute__s like
>> >> __attribute__((packed)).
>> >>
>> >> Example:
>> >> $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
>> >> Defn for struct foo == <struct foo { int bar ; } >
>> >> Hash table occupancy 1/4096 = 0.000244141
>> >> $ echo 'struct __attribute__((packed)) foo { int bar; };' |
>> >> ./scripts/genksyms/genksyms -d
>> >> Hash table occupancy 0/4096 = 0
>> >>
>> >> I assume the __attribute__ part isn't being parsed correctly (looks
>> >> like genksyms is a lex/yacc based C parser).
>> >>
>> >> The issue we have in our out of tree driver (*sadface*) is basically a
>> >> EXPORT_SYMBOL'd function whose signature contains a packed struct.
>> >>
>> >> Theoretically, there should be nothing wrong with exporting a function
>> >> that requires packed structs, and this is just a bug in the lex/yacc
>> >> based parser, right? I assume that not having CONFIG_MODVERSIONS
>> >> coverage of packed structs in particular could lead to potentially
>> >> not-fun bugs? Or is using packed structs in exported function symbols
>> >> with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
>> >> missed?
>> >
>> > Ah, looks like I'm late to the party:
>> > https://lwn.net/Articles/707520/
>>
>> Yeah, would be nice to do something about this.
>
> modversions is ugly, so it would be great if we could dump it.
>
>> IIRC (without re-reading it all), in theory distros would be okay
>> without modversions if they could just provide their own explicit
>> versioning. They take care about ABIs, so they can version things
>> carefully if they had to change.
>
> We have not provided any alternative solution for this, haven't we?
>
> In your patch (https://lwn.net/Articles/707729/),
> you proposed CONFIG_MODULE_ABI_EXPLICIT.
Right, that was just my first proposal, but I am not confident that I
understood everybody's requirements. I don't think the distro people
had much time to to test things out.
One possible shortcoming with that patch is no per-symbol version. The
distro may break an ABI for a security fix, but they don't want to break
all out of tree modules if it's an obscure ABI. The counter argument to
that is they should just rename the symbol in their kernel for such
cases, so I didn't implement it without somebody describing a good
requirement.
> If it is good enough for distros, we merge it first,
> give them time to migrate over to it, then finally remove modversions??
I guess. Do we really need to merge and wait? If they _really_ want it,
and won't put in effort to convert their kernel packaging, then they
can carry the patch and support it quite easily. The code doesn't
change frequently so it should not be a big roadblock
I'm more concerned about developer and hobbyists etc who don't have the
resources. But IIRC we are satisfied that git version has superseded
the benefits of modversions for that case now.
Thanks,
Nick
On Tue, 2019-08-27 at 22:42 +1000, Nicholas Piggin wrote:
> Masahiro Yamada's on August 27, 2019 8:49 pm:
> > Hi.
> >
> > On Tue, Aug 27, 2019 at 6:59 PM Nicholas Piggin <[email protected]> wrote:
> > > Nick Desaulniers's on August 27, 2019 8:57 am:
> > > > On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
> > > > <[email protected]> wrote:
> > > > > I'm looking into a linkage failure for one of our device kernels, and
> > > > > it seems that genksyms isn't producing a hash value correctly for
> > > > > aggregate definitions that contain __attribute__s like
> > > > > __attribute__((packed)).
> > > > >
> > > > > Example:
> > > > > $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
> > > > > Defn for struct foo == <struct foo { int bar ; } >
> > > > > Hash table occupancy 1/4096 = 0.000244141
> > > > > $ echo 'struct __attribute__((packed)) foo { int bar; };' |
> > > > > ./scripts/genksyms/genksyms -d
> > > > > Hash table occupancy 0/4096 = 0
> > > > >
> > > > > I assume the __attribute__ part isn't being parsed correctly (looks
> > > > > like genksyms is a lex/yacc based C parser).
> > > > >
> > > > > The issue we have in our out of tree driver (*sadface*) is basically a
> > > > > EXPORT_SYMBOL'd function whose signature contains a packed struct.
> > > > >
> > > > > Theoretically, there should be nothing wrong with exporting a function
> > > > > that requires packed structs, and this is just a bug in the lex/yacc
> > > > > based parser, right? I assume that not having CONFIG_MODVERSIONS
> > > > > coverage of packed structs in particular could lead to potentially
> > > > > not-fun bugs? Or is using packed structs in exported function symbols
> > > > > with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
> > > > > missed?
> > > >
> > > > Ah, looks like I'm late to the party:
> > > > https://lwn.net/Articles/707520/
> > >
> > > Yeah, would be nice to do something about this.
> >
> > modversions is ugly, so it would be great if we could dump it.
> >
> > > IIRC (without re-reading it all), in theory distros would be okay
> > > without modversions if they could just provide their own explicit
> > > versioning. They take care about ABIs, so they can version things
> > > carefully if they had to change.
Debian doesn't currently have any other way of detecting ABI changes
(other than eyeballing diffs).
I know there have been proposals of using libabigail for this instead,
but I'm not sure how far those progressed.
> > We have not provided any alternative solution for this, haven't we?
> >
> > In your patch (https://lwn.net/Articles/707729/),
> > you proposed CONFIG_MODULE_ABI_EXPLICIT.
>
> Right, that was just my first proposal, but I am not confident that I
> understood everybody's requirements. I don't think the distro people
> had much time to to test things out.
>
> One possible shortcoming with that patch is no per-symbol version. The
> distro may break an ABI for a security fix, but they don't want to break
> all out of tree modules if it's an obscure ABI.
Right, for example the KVM kABI is only meant for in-tree modules (like
kvm_intel) and in Debian we do not change the "ABI version" and require
rebuilding out-of-tree modules just because that ABI changes.
Currently we maintain explicit lists of exported symbols and exporting
modules for which we ignore ABI changes at build time.
> The counter argument to
> that is they should just rename the symbol in their kernel for such
> cases, so I didn't implement it without somebody describing a good
> requirement.
[...]
Sometimes it is just a single function that changes, but often a
structure change can affect large numbers of functions. For example,
if KVM adds a member to an operations struct that can indirectly change
the ABI for most of its exported functions. We wouldn't want to change
the ABI version but would still want to prevent loading mismatched kvm
and kvm_intel versions. It would be a lot more work to change all of
the affected function names.
An alternative to symbol version matching that I think would work for
us is: if a module's exports or imports match the "changes ignored"
list then the module can only be loaded on the exact version of the
kernel, otherwise it only needs to match the ABI version. I think that
would avoid the need for carrying symbol versions, but we would still
need a build-time ABI check and a way of flagging which symbols need
the tighter version match.
Ben.
--
Ben Hutchings
I'm always amazed by the number of people who take up solipsism because
they heard someone else explain it. - E*Borg on alt.fan.pratchett
On Tue, Aug 27, 2019 at 04:34:15PM +0100, Ben Hutchings wrote:
> On Tue, 2019-08-27 at 22:42 +1000, Nicholas Piggin wrote:
> > Masahiro Yamada's on August 27, 2019 8:49 pm:
> > > Hi.
> > >
> > > On Tue, Aug 27, 2019 at 6:59 PM Nicholas Piggin <[email protected]> wrote:
> > > > Nick Desaulniers's on August 27, 2019 8:57 am:
> > > > > On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
> > > > > <[email protected]> wrote:
> > > > > > I'm looking into a linkage failure for one of our device kernels, and
> > > > > > it seems that genksyms isn't producing a hash value correctly for
> > > > > > aggregate definitions that contain __attribute__s like
> > > > > > __attribute__((packed)).
> > > > > >
> > > > > > Example:
> > > > > > $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
> > > > > > Defn for struct foo == <struct foo { int bar ; } >
> > > > > > Hash table occupancy 1/4096 = 0.000244141
> > > > > > $ echo 'struct __attribute__((packed)) foo { int bar; };' |
> > > > > > ./scripts/genksyms/genksyms -d
> > > > > > Hash table occupancy 0/4096 = 0
> > > > > >
> > > > > > I assume the __attribute__ part isn't being parsed correctly (looks
> > > > > > like genksyms is a lex/yacc based C parser).
> > > > > >
> > > > > > The issue we have in our out of tree driver (*sadface*) is basically a
> > > > > > EXPORT_SYMBOL'd function whose signature contains a packed struct.
> > > > > >
> > > > > > Theoretically, there should be nothing wrong with exporting a function
> > > > > > that requires packed structs, and this is just a bug in the lex/yacc
> > > > > > based parser, right? I assume that not having CONFIG_MODVERSIONS
> > > > > > coverage of packed structs in particular could lead to potentially
> > > > > > not-fun bugs? Or is using packed structs in exported function symbols
> > > > > > with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
> > > > > > missed?
> > > > >
> > > > > Ah, looks like I'm late to the party:
> > > > > https://lwn.net/Articles/707520/
> > > >
> > > > Yeah, would be nice to do something about this.
> > >
> > > modversions is ugly, so it would be great if we could dump it.
> > >
> > > > IIRC (without re-reading it all), in theory distros would be okay
> > > > without modversions if they could just provide their own explicit
> > > > versioning. They take care about ABIs, so they can version things
> > > > carefully if they had to change.
>
> Debian doesn't currently have any other way of detecting ABI changes
> (other than eyeballing diffs).
>
> I know there have been proposals of using libabigail for this instead,
> but I'm not sure how far those progressed.
Google has started using libabigail to track api changes in AOSP, here's
a patch that updates the ABI file after changing it:
https://android-review.googlesource.com/c/kernel/common/+/1108662
Note, there are issues with it, and some rough edges, but I think it can
work.
But, it means nothing at module load time, this is only at build-check
time. At least modversions would prevent module loading in some cases.
thanks,
greg k-h
On Tue, 2019-08-27 at 19:09 +0200, Greg KH wrote:
> On Tue, Aug 27, 2019 at 04:34:15PM +0100, Ben Hutchings wrote:
> > On Tue, 2019-08-27 at 22:42 +1000, Nicholas Piggin wrote:
> > > Masahiro Yamada's on August 27, 2019 8:49 pm:
[...]
> > > > modversions is ugly, so it would be great if we could dump it.
> > > >
> > > > > IIRC (without re-reading it all), in theory distros would be okay
> > > > > without modversions if they could just provide their own explicit
> > > > > versioning. They take care about ABIs, so they can version things
> > > > > carefully if they had to change.
> >
> > Debian doesn't currently have any other way of detecting ABI changes
> > (other than eyeballing diffs).
> >
> > I know there have been proposals of using libabigail for this instead,
> > but I'm not sure how far those progressed.
>
> Google has started using libabigail to track api changes in AOSP, here's
> a patch that updates the ABI file after changing it:
> https://android-review.googlesource.com/c/kernel/common/+/1108662
>
> Note, there are issues with it, and some rough edges, but I think it can
> work.
Thanks for the pointer.
> But, it means nothing at module load time, this is only at build-check
> time. At least modversions would prevent module loading in some cases.
Right, but I *think* that would be enough if we could mark modules for
strict (exact version) or loose ("ABI version") matching as I outlined.
Ben.
--
Ben Hutchings
I'm always amazed by the number of people who take up solipsism because
they heard someone else explain it. - E*Borg on alt.fan.pratchett
Ben Hutchings's on August 28, 2019 1:34 am:
> On Tue, 2019-08-27 at 22:42 +1000, Nicholas Piggin wrote:
>> Masahiro Yamada's on August 27, 2019 8:49 pm:
>> > Hi.
>> >
>> > On Tue, Aug 27, 2019 at 6:59 PM Nicholas Piggin <[email protected]> wrote:
>> > > Nick Desaulniers's on August 27, 2019 8:57 am:
>> > > > On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
>> > > > <[email protected]> wrote:
>> > > > > I'm looking into a linkage failure for one of our device kernels, and
>> > > > > it seems that genksyms isn't producing a hash value correctly for
>> > > > > aggregate definitions that contain __attribute__s like
>> > > > > __attribute__((packed)).
>> > > > >
>> > > > > Example:
>> > > > > $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
>> > > > > Defn for struct foo == <struct foo { int bar ; } >
>> > > > > Hash table occupancy 1/4096 = 0.000244141
>> > > > > $ echo 'struct __attribute__((packed)) foo { int bar; };' |
>> > > > > ./scripts/genksyms/genksyms -d
>> > > > > Hash table occupancy 0/4096 = 0
>> > > > >
>> > > > > I assume the __attribute__ part isn't being parsed correctly (looks
>> > > > > like genksyms is a lex/yacc based C parser).
>> > > > >
>> > > > > The issue we have in our out of tree driver (*sadface*) is basically a
>> > > > > EXPORT_SYMBOL'd function whose signature contains a packed struct.
>> > > > >
>> > > > > Theoretically, there should be nothing wrong with exporting a function
>> > > > > that requires packed structs, and this is just a bug in the lex/yacc
>> > > > > based parser, right? I assume that not having CONFIG_MODVERSIONS
>> > > > > coverage of packed structs in particular could lead to potentially
>> > > > > not-fun bugs? Or is using packed structs in exported function symbols
>> > > > > with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
>> > > > > missed?
>> > > >
>> > > > Ah, looks like I'm late to the party:
>> > > > https://lwn.net/Articles/707520/
>> > >
>> > > Yeah, would be nice to do something about this.
>> >
>> > modversions is ugly, so it would be great if we could dump it.
>> >
>> > > IIRC (without re-reading it all), in theory distros would be okay
>> > > without modversions if they could just provide their own explicit
>> > > versioning. They take care about ABIs, so they can version things
>> > > carefully if they had to change.
>
> Debian doesn't currently have any other way of detecting ABI changes
> (other than eyeballing diffs).
>
> I know there have been proposals of using libabigail for this instead,
> but I'm not sure how far those progressed.
>
>> > We have not provided any alternative solution for this, haven't we?
>> >
>> > In your patch (https://lwn.net/Articles/707729/),
>> > you proposed CONFIG_MODULE_ABI_EXPLICIT.
>>
>> Right, that was just my first proposal, but I am not confident that I
>> understood everybody's requirements. I don't think the distro people
>> had much time to to test things out.
>>
>> One possible shortcoming with that patch is no per-symbol version. The
>> distro may break an ABI for a security fix, but they don't want to break
>> all out of tree modules if it's an obscure ABI.
>
> Right, for example the KVM kABI is only meant for in-tree modules (like
> kvm_intel) and in Debian we do not change the "ABI version" and require
> rebuilding out-of-tree modules just because that ABI changes.
> Currently we maintain explicit lists of exported symbols and exporting
> modules for which we ignore ABI changes at build time.
>
>> The counter argument to
>> that is they should just rename the symbol in their kernel for such
>> cases, so I didn't implement it without somebody describing a good
>> requirement.
> [...]
>
> Sometimes it is just a single function that changes, but often a
> structure change can affect large numbers of functions. For example,
> if KVM adds a member to an operations struct that can indirectly change
> the ABI for most of its exported functions. We wouldn't want to change
> the ABI version but would still want to prevent loading mismatched kvm
> and kvm_intel versions. It would be a lot more work to change all of
> the affected function names.
You could change just a single symbol name though :)
> An alternative to symbol version matching that I think would work for
> us is: if a module's exports or imports match the "changes ignored"
> list then the module can only be loaded on the exact version of the
> kernel, otherwise it only needs to match the ABI version. I think that
> would avoid the need for carrying symbol versions, but we would still
> need a build-time ABI check and a way of flagging which symbols need
> the tighter version match.
Just trying to think how best to express that.
[ Aside, the whole symbol name resolution linking stuff does matching on
on any number of ~arbitrary strings that you can generate as you like,
and symbol tables are something that all existing tools and libs
understand.
So I strongly favour using that as the back end for our "version"
resolution system _if at all possible_ rather than adding these extra
bits of crud that really just do the same thing. At least for a first
pass, I don't want to over-engineer things.
Then it hopefully becomes a matter of adding some helper macros and
build facilities on top of that which can contain everyone's
requirements mostly within .config and perhaps a very small patch.
A bit more work with preprocessor macros etc is far preferable to
linking and loading "features" IMO]
Back to your case, is it sufficient to have just an internal and an
external module version where the kernel provides both and your in-tree
modules match on the internal, others match on external?
Thanks,
Nick
On Tue, Aug 27, 2019 at 7:26 PM Nicholas Piggin <[email protected]> wrote:
>
> Ben Hutchings's on August 28, 2019 1:34 am:
> > On Tue, 2019-08-27 at 22:42 +1000, Nicholas Piggin wrote:
> >> Masahiro Yamada's on August 27, 2019 8:49 pm:
> >> > Hi.
> >> >
> >> > On Tue, Aug 27, 2019 at 6:59 PM Nicholas Piggin <[email protected]> wrote:
> >> > > Nick Desaulniers's on August 27, 2019 8:57 am:
> >> > > > On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
> >> > > > <[email protected]> wrote:
> >> > > > > I'm looking into a linkage failure for one of our device kernels, and
> >> > > > > it seems that genksyms isn't producing a hash value correctly for
> >> > > > > aggregate definitions that contain __attribute__s like
> >> > > > > __attribute__((packed)).
> >> > > > >
> >> > > > > Example:
> >> > > > > $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
> >> > > > > Defn for struct foo == <struct foo { int bar ; } >
> >> > > > > Hash table occupancy 1/4096 = 0.000244141
> >> > > > > $ echo 'struct __attribute__((packed)) foo { int bar; };' |
> >> > > > > ./scripts/genksyms/genksyms -d
> >> > > > > Hash table occupancy 0/4096 = 0
> >> > > > >
> >> > > > > I assume the __attribute__ part isn't being parsed correctly (looks
> >> > > > > like genksyms is a lex/yacc based C parser).
> >> > > > >
> >> > > > > The issue we have in our out of tree driver (*sadface*) is basically a
> >> > > > > EXPORT_SYMBOL'd function whose signature contains a packed struct.
> >> > > > >
> >> > > > > Theoretically, there should be nothing wrong with exporting a function
> >> > > > > that requires packed structs, and this is just a bug in the lex/yacc
> >> > > > > based parser, right? I assume that not having CONFIG_MODVERSIONS
> >> > > > > coverage of packed structs in particular could lead to potentially
> >> > > > > not-fun bugs? Or is using packed structs in exported function symbols
> >> > > > > with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
> >> > > > > missed?
> >> > > >
> >> > > > Ah, looks like I'm late to the party:
> >> > > > https://lwn.net/Articles/707520/
> >> > >
> >> > > Yeah, would be nice to do something about this.
> >> >
> >> > modversions is ugly, so it would be great if we could dump it.
> >> >
> >> > > IIRC (without re-reading it all), in theory distros would be okay
> >> > > without modversions if they could just provide their own explicit
> >> > > versioning. They take care about ABIs, so they can version things
> >> > > carefully if they had to change.
> >
> > Debian doesn't currently have any other way of detecting ABI changes
> > (other than eyeballing diffs).
> >
> > I know there have been proposals of using libabigail for this instead,
> > but I'm not sure how far those progressed.
> >
> >> > We have not provided any alternative solution for this, haven't we?
> >> >
> >> > In your patch (https://lwn.net/Articles/707729/),
> >> > you proposed CONFIG_MODULE_ABI_EXPLICIT.
> >>
> >> Right, that was just my first proposal, but I am not confident that I
> >> understood everybody's requirements. I don't think the distro people
> >> had much time to to test things out.
> >>
> >> One possible shortcoming with that patch is no per-symbol version. The
> >> distro may break an ABI for a security fix, but they don't want to break
> >> all out of tree modules if it's an obscure ABI.
> >
> > Right, for example the KVM kABI is only meant for in-tree modules (like
> > kvm_intel) and in Debian we do not change the "ABI version" and require
> > rebuilding out-of-tree modules just because that ABI changes.
> > Currently we maintain explicit lists of exported symbols and exporting
> > modules for which we ignore ABI changes at build time.
> >
> >> The counter argument to
> >> that is they should just rename the symbol in their kernel for such
> >> cases, so I didn't implement it without somebody describing a good
> >> requirement.
> > [...]
> >
> > Sometimes it is just a single function that changes, but often a
> > structure change can affect large numbers of functions. For example,
> > if KVM adds a member to an operations struct that can indirectly change
> > the ABI for most of its exported functions. We wouldn't want to change
> > the ABI version but would still want to prevent loading mismatched kvm
> > and kvm_intel versions. It would be a lot more work to change all of
> > the affected function names.
>
> You could change just a single symbol name though :)
>
> > An alternative to symbol version matching that I think would work for
> > us is: if a module's exports or imports match the "changes ignored"
> > list then the module can only be loaded on the exact version of the
> > kernel, otherwise it only needs to match the ABI version. I think that
> > would avoid the need for carrying symbol versions, but we would still
> > need a build-time ABI check and a way of flagging which symbols need
> > the tighter version match.
>
> Just trying to think how best to express that.
>
> [ Aside, the whole symbol name resolution linking stuff does matching on
> on any number of ~arbitrary strings that you can generate as you like,
> and symbol tables are something that all existing tools and libs
> understand.
>
> So I strongly favour using that as the back end for our "version"
> resolution system _if at all possible_ rather than adding these extra
> bits of crud that really just do the same thing. At least for a first
> pass, I don't want to over-engineer things.
>
> Then it hopefully becomes a matter of adding some helper macros and
> build facilities on top of that which can contain everyone's
> requirements mostly within .config and perhaps a very small patch.
> A bit more work with preprocessor macros etc is far preferable to
> linking and loading "features" IMO]
>
> Back to your case, is it sufficient to have just an internal and an
> external module version where the kernel provides both and your in-tree
> modules match on the internal, others match on external?
>
> Thanks,
> Nick
>
+ some Android folks who are looking into libabigail
(root thread for context:
https://lore.kernel.org/lkml/CAKwvOdnJAApaUhTQs7w_VjSeYBQa0c-TNxRB4xPLi0Y0sOQMMQ@mail.gmail.com/)
I'm only roughly aware of their issue/work, but hopefully they can
collaborate and add their insights to a list of requirements we could
collect, for any kind of replacement to CONFIG_MODVERSIONS/genksyms.
I'm not sure if they were planning on presenting this work at Linux
Plumber's Conference coming up or not, but I hope so. Hopefully that
work will be a general solution for all kernel developers and distros,
not just Android.
This issue is now blocking my compiler upgrade, so I'm going to focus
on just fixing the yacc based parser to understand __attribute__'s on
struct/enum/union declarations.
Speaking with Arnd about this issue, he came up with the test case:
$ echo 'struct __attribute__((packed)) foo { char bar; int baz; };' |
./scripts/genksyms/genksyms -d
Hash table occupancy 0/4096 = 0
$ echo 'struct __attribute__((packed)) foo { char bar; };' |
./scripts/genksyms/genksyms -d
Hash table occupancy 0/4096 = 0
The implication being that because attributes don't parse correctly,
these symbols get a hash value of 0x00 in Module.symvers, meaning we
currently can't detect size changes of packed structs (or any other
attribute-qualified type) and happily proceed loading modules with
different ABIs than the kernel.
I'll probably add some kind of script for unit tests for genksyms,
too. Feels like overkill for something known to be broken where
active replacements are being discussed, but it's blocking my compiler
upgrade for our products, so I'm just going to fix it for now.
--
Thanks,
~Nick Desaulniers
On Wed, Aug 28, 2019 at 10:17:27AM -0700, Nick Desaulniers wrote:
>On Tue, Aug 27, 2019 at 7:26 PM Nicholas Piggin <[email protected]> wrote:
>>
>> Ben Hutchings's on August 28, 2019 1:34 am:
>> > On Tue, 2019-08-27 at 22:42 +1000, Nicholas Piggin wrote:
>> >> Masahiro Yamada's on August 27, 2019 8:49 pm:
>> >> > Hi.
>> >> >
>> >> > On Tue, Aug 27, 2019 at 6:59 PM Nicholas Piggin <[email protected]> wrote:
>> >> > > Nick Desaulniers's on August 27, 2019 8:57 am:
>> >> > > > On Mon, Aug 26, 2019 at 2:22 PM Nick Desaulniers
>> >> > > > <[email protected]> wrote:
>> >> > > > > I'm looking into a linkage failure for one of our device kernels, and
>> >> > > > > it seems that genksyms isn't producing a hash value correctly for
>> >> > > > > aggregate definitions that contain __attribute__s like
>> >> > > > > __attribute__((packed)).
>> >> > > > >
>> >> > > > > Example:
>> >> > > > > $ echo 'struct foo { int bar; };' | ./scripts/genksyms/genksyms -d
>> >> > > > > Defn for struct foo == <struct foo { int bar ; } >
>> >> > > > > Hash table occupancy 1/4096 = 0.000244141
>> >> > > > > $ echo 'struct __attribute__((packed)) foo { int bar; };' |
>> >> > > > > ./scripts/genksyms/genksyms -d
>> >> > > > > Hash table occupancy 0/4096 = 0
>> >> > > > >
>> >> > > > > I assume the __attribute__ part isn't being parsed correctly (looks
>> >> > > > > like genksyms is a lex/yacc based C parser).
>> >> > > > >
>> >> > > > > The issue we have in our out of tree driver (*sadface*) is basically a
>> >> > > > > EXPORT_SYMBOL'd function whose signature contains a packed struct.
>> >> > > > >
>> >> > > > > Theoretically, there should be nothing wrong with exporting a function
>> >> > > > > that requires packed structs, and this is just a bug in the lex/yacc
>> >> > > > > based parser, right? I assume that not having CONFIG_MODVERSIONS
>> >> > > > > coverage of packed structs in particular could lead to potentially
>> >> > > > > not-fun bugs? Or is using packed structs in exported function symbols
>> >> > > > > with CONFIG_MODVERSIONS forbidden in some documentation somewhere I
>> >> > > > > missed?
>> >> > > >
>> >> > > > Ah, looks like I'm late to the party:
>> >> > > > https://lwn.net/Articles/707520/
>> >> > >
>> >> > > Yeah, would be nice to do something about this.
>> >> >
>> >> > modversions is ugly, so it would be great if we could dump it.
>> >> >
>> >> > > IIRC (without re-reading it all), in theory distros would be okay
>> >> > > without modversions if they could just provide their own explicit
>> >> > > versioning. They take care about ABIs, so they can version things
>> >> > > carefully if they had to change.
>> >
>> > Debian doesn't currently have any other way of detecting ABI changes
>> > (other than eyeballing diffs).
>> >
>> > I know there have been proposals of using libabigail for this instead,
>> > but I'm not sure how far those progressed.
>> >
>> >> > We have not provided any alternative solution for this, haven't we?
>> >> >
>> >> > In your patch (https://lwn.net/Articles/707729/),
>> >> > you proposed CONFIG_MODULE_ABI_EXPLICIT.
>> >>
>> >> Right, that was just my first proposal, but I am not confident that I
>> >> understood everybody's requirements. I don't think the distro people
>> >> had much time to to test things out.
>> >>
>> >> One possible shortcoming with that patch is no per-symbol version. The
>> >> distro may break an ABI for a security fix, but they don't want to break
>> >> all out of tree modules if it's an obscure ABI.
>> >
>> > Right, for example the KVM kABI is only meant for in-tree modules (like
>> > kvm_intel) and in Debian we do not change the "ABI version" and require
>> > rebuilding out-of-tree modules just because that ABI changes.
>> > Currently we maintain explicit lists of exported symbols and exporting
>> > modules for which we ignore ABI changes at build time.
>> >
>> >> The counter argument to
>> >> that is they should just rename the symbol in their kernel for such
>> >> cases, so I didn't implement it without somebody describing a good
>> >> requirement.
>> > [...]
>> >
>> > Sometimes it is just a single function that changes, but often a
>> > structure change can affect large numbers of functions. For example,
>> > if KVM adds a member to an operations struct that can indirectly change
>> > the ABI for most of its exported functions. We wouldn't want to change
>> > the ABI version but would still want to prevent loading mismatched kvm
>> > and kvm_intel versions. It would be a lot more work to change all of
>> > the affected function names.
>>
>> You could change just a single symbol name though :)
>>
>> > An alternative to symbol version matching that I think would work for
>> > us is: if a module's exports or imports match the "changes ignored"
>> > list then the module can only be loaded on the exact version of the
>> > kernel, otherwise it only needs to match the ABI version. I think that
>> > would avoid the need for carrying symbol versions, but we would still
>> > need a build-time ABI check and a way of flagging which symbols need
>> > the tighter version match.
>>
>> Just trying to think how best to express that.
>>
>> [ Aside, the whole symbol name resolution linking stuff does matching on
>> on any number of ~arbitrary strings that you can generate as you like,
>> and symbol tables are something that all existing tools and libs
>> understand.
>>
>> So I strongly favour using that as the back end for our "version"
>> resolution system _if at all possible_ rather than adding these extra
>> bits of crud that really just do the same thing. At least for a first
>> pass, I don't want to over-engineer things.
>>
>> Then it hopefully becomes a matter of adding some helper macros and
>> build facilities on top of that which can contain everyone's
>> requirements mostly within .config and perhaps a very small patch.
>> A bit more work with preprocessor macros etc is far preferable to
>> linking and loading "features" IMO]
>>
>> Back to your case, is it sufficient to have just an internal and an
>> external module version where the kernel provides both and your in-tree
>> modules match on the internal, others match on external?
>>
>> Thanks,
>> Nick
>>
>
>+ some Android folks who are looking into libabigail
>
>(root thread for context:
>https://lore.kernel.org/lkml/CAKwvOdnJAApaUhTQs7w_VjSeYBQa0c-TNxRB4xPLi0Y0sOQMMQ@mail.gmail.com/)
>
>I'm only roughly aware of their issue/work, but hopefully they can
>collaborate and add their insights to a list of requirements we could
>collect, for any kind of replacement to CONFIG_MODVERSIONS/genksyms.
>I'm not sure if they were planning on presenting this work at Linux
>Plumber's Conference coming up or not, but I hope so. Hopefully that
>work will be a general solution for all kernel developers and distros,
>not just Android.
Several people are currently working on improving libabigail's
capabilities to deal with the kernel binaries. That ranges from fixes
for newer kernel versions or fixes for binaries built with clang to
performance improvements to better integrate the tool into build
processes etc. There is still some work to do to be an actual
replacement for what is used right now to detect ABI incompatibilities.
But things look promising. The work that is done there is supposed to
address general issues, not just Android. In particular, libabigail
should be able to reliably keep track of the observable in-kernel ABI of
the kernel binaries to detect potential incompatibilities with
previously released kernel binaries. Please note, that this all happens
after the build by inspecting binaries. As opposed to other solutions,
in its current form, libabigail does not introduce any checks at
runtime (i.e. module load time).
There are discussions about this topic scheduled at Plumber's in the
Distro MC and the Android MC.
Cheers,
Matthias
>This issue is now blocking my compiler upgrade, so I'm going to focus
>on just fixing the yacc based parser to understand __attribute__'s on
>struct/enum/union declarations.
>
>Speaking with Arnd about this issue, he came up with the test case:
>$ echo 'struct __attribute__((packed)) foo { char bar; int baz; };' |
>./scripts/genksyms/genksyms -d
>Hash table occupancy 0/4096 = 0
>$ echo 'struct __attribute__((packed)) foo { char bar; };' |
>./scripts/genksyms/genksyms -d
>Hash table occupancy 0/4096 = 0
>
>The implication being that because attributes don't parse correctly,
>these symbols get a hash value of 0x00 in Module.symvers, meaning we
>currently can't detect size changes of packed structs (or any other
>attribute-qualified type) and happily proceed loading modules with
>different ABIs than the kernel.
>
>I'll probably add some kind of script for unit tests for genksyms,
>too. Feels like overkill for something known to be broken where
>active replacements are being discussed, but it's blocking my compiler
>upgrade for our products, so I'm just going to fix it for now.
On Sun, Jul 26, 2020 at 4:31 PM Rob Simmons <[email protected]> wrote:
You may have forgotten to send a body; but yes genksysms is brittle
and upstream development on it stopped years ago. Put an attribute
someplace unexpected and the parser will silently fail, making the
modversions hash all zeros, which doesn't protect you from loading a
potentially harmful module with different function signature. If
you're planning on attending plumbers, I plan to put together a brief
talk on the subject.
--
Thanks,
~Nick Desaulniers