ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Helmut Grohne]

Hi,

During a compile test of v4.20-rc7 for armhf (cross build), I
encountered:

ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

Some searching brought up c5157101e77 ("arm64: mm: Export
__sync_icache_dcache() for xen-privcmd"), which fixes a very similar
problem for arm64. Is that commit maybe also needed for arm?

Helmut

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Russell King - ARM Linux]

On Tue, Dec 18, 2018 at 10:36:32AM +0100, Helmut Grohne wrote:
> Hi,
> 
> During a compile test of v4.20-rc7 for armhf (cross build), I
> encountered:
> 
> ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!
> 
> Some searching brought up c5157101e77 ("arm64: mm: Export
> __sync_icache_dcache() for xen-privcmd"), which fixes a very similar
> problem for arm64. Is that commit maybe also needed for arm?

I would rather not export that function for drivers to abuse.

-- 
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
According to speedtest.net: 11.9Mbps down 500kbps up

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Catalin Marinas]

On Tue, Dec 18, 2018 at 10:09:09AM +0000, Russell King wrote:
> On Tue, Dec 18, 2018 at 10:36:32AM +0100, Helmut Grohne wrote:
> > During a compile test of v4.20-rc7 for armhf (cross build), I
> > encountered:
> > 
> > ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!
> > 
> > Some searching brought up c5157101e77 ("arm64: mm: Export
> > __sync_icache_dcache() for xen-privcmd"), which fixes a very similar
> > problem for arm64. Is that commit maybe also needed for arm?
> 
> I would rather not export that function for drivers to abuse.

I think that's a lower risk than the other cache maintenance functions.
This function only deals with the cache maintenance to PoU while
drivers, in general, are more interested in the PoC.

-- 
Catalin

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Boris Ostrovsky]

On 12/18/18 8:21 AM, Catalin Marinas wrote:
> On Tue, Dec 18, 2018 at 10:09:09AM +0000, Russell King wrote:
>> On Tue, Dec 18, 2018 at 10:36:32AM +0100, Helmut Grohne wrote:
>>> During a compile test of v4.20-rc7 for armhf (cross build), I
>>> encountered:
>>>
>>> ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!
>>>
>>> Some searching brought up c5157101e77 ("arm64: mm: Export
>>> __sync_icache_dcache() for xen-privcmd"), which fixes a very similar
>>> problem for arm64. Is that commit maybe also needed for arm?
>> I would rather not export that function for drivers to abuse.


(I am reviving this thread because Arnd submitted a patch to work around
this issue
https://lore.kernel.org/lkml/20190304204826.2414365-1-arnd@arndb.de/ )

Even when the same routine in arm64 is exported for this very reason?

The patch above it going into a lot of trouble to make things work so I
am hoping this can be re-considered.


Thanks.
-boris


> I think that's a lower risk than the other cache maintenance functions.
> This function only deals with the cache maintenance to PoU while
> drivers, in general, are more interested in the PoC.
>




_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Catalin Marinas]

On Mon, Mar 04, 2019 at 05:04:10PM -0500, Boris Ostrovsky wrote:
> On 12/18/18 8:21 AM, Catalin Marinas wrote:
> > On Tue, Dec 18, 2018 at 10:09:09AM +0000, Russell King wrote:
> >> On Tue, Dec 18, 2018 at 10:36:32AM +0100, Helmut Grohne wrote:
> >>> During a compile test of v4.20-rc7 for armhf (cross build), I
> >>> encountered:
> >>>
> >>> ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!
> >>>
> >>> Some searching brought up c5157101e77 ("arm64: mm: Export
> >>> __sync_icache_dcache() for xen-privcmd"), which fixes a very similar
> >>> problem for arm64. Is that commit maybe also needed for arm?
> >> I would rather not export that function for drivers to abuse.
> 
> (I am reviving this thread because Arnd submitted a patch to work around
> this issue
> https://lore.kernel.org/lkml/20190304204826.2414365-1-arnd@arndb.de/ )
> 
> Even when the same routine in arm64 is exported for this very reason?
> 
> The patch above it going into a lot of trouble to make things work so I
> am hoping this can be re-considered.

It's Russell's decision whether to export this or not for arch/arm/. As
I said, I don't really see drivers abusing this since it doesn't help
with DMA cache maintenance (it only cleans to the PoU rather than PoC as
needed for DMA).

Now, the actual question is whether set_pte_at() is allowed to be called
from loadable modules. If yes, then __sync_icache_dcache() needs to be
exported. Otherwise, Arnd's patch looks fine. Grep'ing (5.0-rc3)
drivers/ for set_pte_at() I can only see 4 instances: one in
gpu/drm/i915/ which is not compiled as a module and three in xen/. Of
the latter, one is x86 only, the second is built-in and the third is
fixed by Arnd's patch.

-- 
Catalin

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Boris Ostrovsky]

On 3/4/19 5:28 PM, Catalin Marinas wrote:
> On Mon, Mar 04, 2019 at 05:04:10PM -0500, Boris Ostrovsky wrote:
>> On 12/18/18 8:21 AM, Catalin Marinas wrote:
>>> On Tue, Dec 18, 2018 at 10:09:09AM +0000, Russell King wrote:
>>>> On Tue, Dec 18, 2018 at 10:36:32AM +0100, Helmut Grohne wrote:
>>>>> During a compile test of v4.20-rc7 for armhf (cross build), I
>>>>> encountered:
>>>>>
>>>>> ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!
>>>>>
>>>>> Some searching brought up c5157101e77 ("arm64: mm: Export
>>>>> __sync_icache_dcache() for xen-privcmd"), which fixes a very similar
>>>>> problem for arm64. Is that commit maybe also needed for arm?
>>>> I would rather not export that function for drivers to abuse.
>> (I am reviving this thread because Arnd submitted a patch to work around
>> this issue
>> https://lore.kernel.org/lkml/20190304204826.2414365-1-arnd@arndb.de/ )
>>
>> Even when the same routine in arm64 is exported for this very reason?
>>
>> The patch above it going into a lot of trouble to make things work so I
>> am hoping this can be re-considered.
> It's Russell's decision whether to export this or not for arch/arm/. 

OK, I didn't realize there are different maintainers for arm and arm64.

In any case, the main reason I am asking this question is because the
two architectures have different policies for (what appears to me) a
very similar interface. If __sync_icache_dcache() is kept private in
arm, should then c5157101e77 be reverted (assuming Arnd's patch goes in)?

-boris

> As
> I said, I don't really see drivers abusing this since it doesn't help
> with DMA cache maintenance (it only cleans to the PoU rather than PoC as
> needed for DMA).
>
> Now, the actual question is whether set_pte_at() is allowed to be called
> from loadable modules. If yes, then __sync_icache_dcache() needs to be
> exported. Otherwise, Arnd's patch looks fine. Grep'ing (5.0-rc3)
> drivers/ for set_pte_at() I can only see 4 instances: one in
> gpu/drm/i915/ which is not compiled as a module and three in xen/. Of
> the latter, one is x86 only, the second is built-in and the third is
> fixed by Arnd's patch.
>


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Arnd Bergmann]

On Mon, Mar 4, 2019 at 11:49 PM Boris Ostrovsky
<boris.ostrovsky@oracle.com> wrote:
> On 3/4/19 5:28 PM, Catalin Marinas wrote:
> > On Mon, Mar 04, 2019 at 05:04:10PM -0500, Boris Ostrovsky wrote:
> >> On 12/18/18 8:21 AM, Catalin Marinas wrote:
> >>> On Tue, Dec 18, 2018 at 10:09:09AM +0000, Russell King wrote:
> >>>> On Tue, Dec 18, 2018 at 10:36:32AM +0100, Helmut Grohne wrote:
> >>>>> During a compile test of v4.20-rc7 for armhf (cross build), I
> >>>>> encountered:
> >>>>>
> >>>>> ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!
> >>>>>
> >>>>> Some searching brought up c5157101e77 ("arm64: mm: Export
> >>>>> __sync_icache_dcache() for xen-privcmd"), which fixes a very similar
> >>>>> problem for arm64. Is that commit maybe also needed for arm?
> >>>> I would rather not export that function for drivers to abuse.
> >> (I am reviving this thread because Arnd submitted a patch to work around
> >> this issue
> >> https://lore.kernel.org/lkml/20190304204826.2414365-1-arnd@arndb.de/ )
> >>
> >> Even when the same routine in arm64 is exported for this very reason?
> >>
> >> The patch above it going into a lot of trouble to make things work so I
> >> am hoping this can be re-considered.
> > It's Russell's decision whether to export this or not for arch/arm/.
>
> OK, I didn't realize there are different maintainers for arm and arm64.
>
> In any case, the main reason I am asking this question is because the
> two architectures have different policies for (what appears to me) a
> very similar interface. If __sync_icache_dcache() is kept private in
> arm, should then c5157101e77 be reverted (assuming Arnd's patch goes in)?

I don't think it make much of a difference. All callers are in the arch/, mm/,
drivers/xen, and fs/dax, and this is the only one that can currently be
a module, so I don't think we would need it otherwise.

       Arnd

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!

[Ben Hutchings]

[-- Attachment #1.1: Type: text/plain, Size: 651 bytes --]

On Tue, 2018-12-18 at 10:36 +0100, Helmut Grohne wrote:
> Hi,
> 
> During a compile test of v4.20-rc7 for armhf (cross build), I
> encountered:
> 
> ERROR: "__sync_icache_dcache" [drivers/xen/xen-privcmd.ko] undefined!
> 
> Some searching brought up c5157101e77 ("arm64: mm: Export
> __sync_icache_dcache() for xen-privcmd"), which fixes a very similar
> problem for arm64. Is that commit maybe also needed for arm?

Yes, I submitted the corresponding patch some months ago:
<https://marc.info/?l=linux-arm-kernel&m=153134944429241>

Ben.

> Helmut
-- 
Ben Hutchings
Anthony's Law of Force: Don't force it, get a larger hammer.


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 176 bytes --]

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel