Quantcast

NAK Re: [SRU][Zesty] Support SMMU passthrough using the default domain

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

NAK Re: [SRU][Zesty] Support SMMU passthrough using the default domain

Paolo Pisati-5
On Tue, May 16, 2017 at 10:11 PM, Manoj Iyer <[hidden email]> wrote:

> This patch series adds support for SMMU passthrough using the default domain,
> the patch series implements a kernel command-line option iommu.passthrough=0/1
> 1 - Bypass the IOMMU for DMA, 0 - Use IOMMU translation for DMA. This feature
> is desired on both Qualcomm Centriq 2400 and Thunder-X platforms.
>
> The bug report http://bugs.launchpad.net/bugs/1688158 has detailed test results
> from Qualcomm demonstrating performance improvements using passthrough. I have
> also tested the kernel on Thundex-X for any regressions, for both boot test and
> pci passthrough.
>
> Please review and consider for SRU.

Doesn't build.
--
bye,
p.

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: NAK Re: [SRU][Zesty] Support SMMU passthrough using the default domain

Manoj Iyer


On Thu, May 18, 2017 at 4:24 AM, Paolo Pisati <[hidden email]> wrote:
On Tue, May 16, 2017 at 10:11 PM, Manoj Iyer <[hidden email]> wrote:
This patch series adds support for SMMU passthrough using the default domain, the patch series implements a kernel command-line option iommu.passthrough=0/1 1 - Bypass the IOMMU for DMA, 0 - Use IOMMU translation for DMA. This feature is desired on both Qualcomm Centriq 2400 and Thunder-X platforms. The bug report http://bugs.launchpad.net/bugs/1688158 has detailed test results from Qualcomm demonstrating performance improvements using passthrough. I have also tested the kernel on Thundex-X for any regressions, for both boot test and pci passthrough. Please review and consider for SRU.
Doesn't build.

Could you give me a bit more details? The patches were picked from linux-next and built for testing.

--
bye, p.

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: NAK Re: [SRU][Zesty] Support SMMU passthrough using the default domain

Paolo Pisati-5
I've applied all your patches on top of zesty/master [1], and in a
zesty amd64 chroot i ran:

$ fdr clean && debian/rules build

...
  CC      drivers/iommu/intel-iommu.o
/home/ppisati/ubuntu/drivers/iommu/intel-iommu.c:5393:31: error:
static declaration of 'intel_iommu_ops' follows non-static declaration
 static const struct iommu_ops intel_iommu_ops = {
                               ^~~~~~~~~~~~~~~
/home/ppisati/ubuntu/drivers/iommu/intel-iommu.c:551:24: note:
previous declaration of 'intel_iommu_ops' was here
 const struct iommu_ops intel_iommu_ops;
                        ^~~~~~~~~~~~~~~
/home/ppisati/ubuntu/scripts/Makefile.build:294: recipe for target
'drivers/iommu/intel-iommu.o' failed
make[4]: *** [drivers/iommu/intel-iommu.o] Error 1
/home/ppisati/ubuntu/scripts/Makefile.build:567: recipe for target
'drivers/iommu' failed
make[3]: *** [drivers/iommu] Error 2
make[3]: *** Waiting for unfinished jobs....
/home/ppisati/ubuntu/Makefile:1020: recipe for target 'drivers' failed
make[2]: *** [drivers] Error 2
make[2]: Leaving directory '/home/ppisati/ubuntu/debian/build/build-generic'
Makefile:150: recipe for target 'sub-make' failed
make[1]: *** [sub-make] Error 2
make[1]: Leaving directory '/home/ppisati/ubuntu'
debian/rules.d/2-binary-arch.mk:70: recipe for target
'/home/ppisati/ubuntu/debian/stamps/stamp-build-generic' failed
make: *** [/home/ppisati/ubuntu/debian/stamps/stamp-build-generic] Error 2


1: https://git.launchpad.net/~p-pisati/ubuntu/+source/linux/log/?h=z-master-lp1688158

On Thu, May 18, 2017 at 5:10 PM, Manoj Iyer <[hidden email]> wrote:

>
>
> On Thu, May 18, 2017 at 4:24 AM, Paolo Pisati <[hidden email]>
> wrote:
>
> On Tue, May 16, 2017 at 10:11 PM, Manoj Iyer <[hidden email]>
> wrote:
>
> This patch series adds support for SMMU passthrough using the default
> domain, the patch series implements a kernel command-line option
> iommu.passthrough=0/1 1 - Bypass the IOMMU for DMA, 0 - Use IOMMU
> translation for DMA. This feature is desired on both Qualcomm Centriq 2400
> and Thunder-X platforms. The bug report
> http://bugs.launchpad.net/bugs/1688158 has detailed test results from
> Qualcomm demonstrating performance improvements using passthrough. I have
> also tested the kernel on Thundex-X for any regressions, for both boot test
> and pci passthrough. Please review and consider for SRU.
>
> Doesn't build.
>
>
> Could you give me a bit more details? The patches were picked from
> linux-next and built for testing.
>
> --
> bye, p.



--
bye,
p.

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: NAK Re: [SRU][Zesty] Support SMMU passthrough using the default domain

Manoj Iyer


On Thu, May 18, 2017 at 10:34 AM, Paolo Pisati <[hidden email]> wrote:
I've applied all your patches on top of zesty/master [1], and in a zesty amd64 chroot i ran: $ fdr clean && debian/rules build ... CC drivers/iommu/intel-iommu.o /home/ppisati/ubuntu/drivers/iommu/intel-iommu.c:5393:31: error: static declaration of 'intel_iommu_ops' follows non-static declaration static const struct iommu_ops intel_iommu_ops = { ^~~~~~~~~~~~~~~ /home/ppisati/ubuntu/drivers/iommu/intel-iommu.c:551:24: note: previous declaration of 'intel_iommu_ops' was here const struct iommu_ops intel_iommu_ops; ^~~~~~~~~~~~~~~ /home/ppisati/ubuntu/scripts/Makefile.build:294: recipe for target 'drivers/iommu/intel-iommu.o' failed make[4]: *** [drivers/iommu/intel-iommu.o] Error 1 /home/ppisati/ubuntu/scripts/Makefile.build:567: recipe for target 'drivers/iommu' failed make[3]: *** [drivers/iommu] Error 2 make[3]: *** Waiting for unfinished jobs.... /home/ppisati/ubuntu/Makefile:1020: recipe for target 'drivers' failed make[2]: *** [drivers] Error 2 make[2]: Leaving directory '/home/ppisati/ubuntu/debian/build/build-generic' Makefile:150: recipe for target 'sub-make' failed make[1]: *** [sub-make] Error 2 make[1]: Leaving directory '/home/ppisati/ubuntu' debian/rules.d/2-binary-arch.mk:70: recipe for target '/home/ppisati/ubuntu/debian/stamps/stamp-build-generic' failed make: *** [/home/ppisati/ubuntu/debian/stamps/stamp-build-generic] Error 2

Ouch! let me take a look at a that and resend these patches. 

1: https://git.launchpad.net/~p-pisati/ubuntu/+source/linux/log/?h=z-master-lp1688158 On Thu, May 18, 2017 at 5:10 PM, Manoj Iyer <[hidden email]> wrote:
On Thu, May 18, 2017 at 4:24 AM, Paolo Pisati <[hidden email]> wrote: On Tue, May 16, 2017 at 10:11 PM, Manoj Iyer <[hidden email]> wrote: This patch series adds support for SMMU passthrough using the default domain, the patch series implements a kernel command-line option iommu.passthrough=0/1 1 - Bypass the IOMMU for DMA, 0 - Use IOMMU translation for DMA. This feature is desired on both Qualcomm Centriq 2400 and Thunder-X platforms. The bug report http://bugs.launchpad.net/bugs/1688158 has detailed test results from Qualcomm demonstrating performance improvements using passthrough. I have also tested the kernel on Thundex-X for any regressions, for both boot test and pci passthrough. Please review and consider for SRU. Doesn't build. Could you give me a bit more details? The patches were picked from linux-next and built for testing. -- bye, p.
--
bye, p.

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: NAK Re: [SRU][Zesty] Support SMMU passthrough using the default domain

Manoj Iyer
I figured out what the problem is ... I missed one patch in my list of patches I sent you. To avoid such mistakes with large patch set I should have used pull requests. Let me send you a pull request for this patch series so that I don't mess it up when I do a git send-email. 

On Thu, May 18, 2017 at 10:44 AM, Manoj Iyer <[hidden email]> wrote:


On Thu, May 18, 2017 at 10:34 AM, Paolo Pisati <[hidden email]> wrote:
I've applied all your patches on top of zesty/master [1], and in a zesty amd64 chroot i ran: $ fdr clean && debian/rules build ... CC drivers/iommu/intel-iommu.o /home/ppisati/ubuntu/drivers/iommu/intel-iommu.c:5393:31: error: static declaration of 'intel_iommu_ops' follows non-static declaration static const struct iommu_ops intel_iommu_ops = { ^~~~~~~~~~~~~~~ /home/ppisati/ubuntu/drivers/iommu/intel-iommu.c:551:24: note: previous declaration of 'intel_iommu_ops' was here const struct iommu_ops intel_iommu_ops; ^~~~~~~~~~~~~~~ /home/ppisati/ubuntu/scripts/Makefile.build:294: recipe for target 'drivers/iommu/intel-iommu.o' failed make[4]: *** [drivers/iommu/intel-iommu.o] Error 1 /home/ppisati/ubuntu/scripts/Makefile.build:567: recipe for target 'drivers/iommu' failed make[3]: *** [drivers/iommu] Error 2 make[3]: *** Waiting for unfinished jobs.... /home/ppisati/ubuntu/Makefile:1020: recipe for target 'drivers' failed make[2]: *** [drivers] Error 2 make[2]: Leaving directory '/home/ppisati/ubuntu/debian/build/build-generic' Makefile:150: recipe for target 'sub-make' failed make[1]: *** [sub-make] Error 2 make[1]: Leaving directory '/home/ppisati/ubuntu' debian/rules.d/2-binary-arch.mk:70: recipe for target '/home/ppisati/ubuntu/debian/stamps/stamp-build-generic' failed make: *** [/home/ppisati/ubuntu/debian/stamps/stamp-build-generic] Error 2

Ouch! let me take a look at a that and resend these patches. 

1: https://git.launchpad.net/~p-pisati/ubuntu/+source/linux/log/?h=z-master-lp1688158 On Thu, May 18, 2017 at 5:10 PM, Manoj Iyer <[hidden email]> wrote:
On Thu, May 18, 2017 at 4:24 AM, Paolo Pisati <[hidden email]> wrote: On Tue, May 16, 2017 at 10:11 PM, Manoj Iyer <[hidden email]> wrote: This patch series adds support for SMMU passthrough using the default domain, the patch series implements a kernel command-line option iommu.passthrough=0/1 1 - Bypass the IOMMU for DMA, 0 - Use IOMMU translation for DMA. This feature is desired on both Qualcomm Centriq 2400 and Thunder-X platforms. The bug report http://bugs.launchpad.net/bugs/1688158 has detailed test results from Qualcomm demonstrating performance improvements using passthrough. I have also tested the kernel on Thundex-X for any regressions, for both boot test and pci passthrough. Please review and consider for SRU. Doesn't build. Could you give me a bit more details? The patches were picked from linux-next and built for testing. -- bye, p.
--
bye, p.

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Loading...