Hi Mark, Alex,
I would like to make a LSK v3.10 request from ARM to support our partners using 3.10 with Android and MPG drivers. Those drivers require sMMU support in the kernel to initially support handling of fragmented device memory.
So the request is to back port the driver from a stable place, so I suggest the v3.18 release version - https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drive...
This driver is made up of many individual commits - so I'm not sure how you want to handle it, so have not listed them here, but obviously would be happy to do so.
Will, the driver author, is on cc and would be happy to give advice regarding dependencies, etc. But he has other work and so can't take a big support burden here.
I'm assuming this needs to go to TSC for approval - so I've cc'ed Roger and Mark.
Thanks, James
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
Hi James,
Do you know if there's any available systems which could be used for testing this, or users willing to collaborate on that? I've looked at current mainline and I'm not seeing any references to the compatible strings outside of the drivers and their binding documents.
Thanks, Mark
On 16 January 2015 at 10:19, James King James.King@arm.com wrote:
Hi Mark, Alex,
I would like to make a LSK v3.10 request from ARM to support our partners using 3.10 with Android and MPG drivers. Those drivers require sMMU support in the kernel to initially support handling of fragmented device memory.
So the request is to back port the driver from a stable place, so I suggest the v3.18 release version - https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drive...
This driver is made up of many individual commits – so I’m not sure how you want to handle it, so have not listed them here, but obviously would be happy to do so.
Will, the driver author, is on cc and would be happy to give advice regarding dependencies, etc. But he has other work and so can’t take a big support burden here.
I’m assuming this needs to go to TSC for approval – so I’ve cc’ed Roger and Mark.
Thanks, James
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
Hi Mark,
Any Juno board should be fine as that has a SMMUv1. Once in LSK, we can test it is still ok on a v2 part.
Will – are there some tests (LTP?) that can be used to test it is working ok when back ported?
Thanks, James
From: Mark Brown [mailto:broonie@linaro.org] Sent: 19 January 2015 16:16 To: James King Cc: alex.shi@linaro.org; linaro-kernel@lists.linaro.org; Mark Hambleton; Will Deacon; Roger Teague Subject: Re: 3.10 LSK request
Hi James,
Do you know if there's any available systems which could be used for testing this, or users willing to collaborate on that? I've looked at current mainline and I'm not seeing any references to the compatible strings outside of the drivers and their binding documents.
Thanks, Mark
On 16 January 2015 at 10:19, James King <James.King@arm.commailto:James.King@arm.com> wrote: Hi Mark, Alex,
I would like to make a LSK v3.10 request from ARM to support our partners using 3.10 with Android and MPG drivers. Those drivers require sMMU support in the kernel to initially support handling of fragmented device memory.
So the request is to back port the driver from a stable place, so I suggest the v3.18 release version - https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drive...
This driver is made up of many individual commits – so I’m not sure how you want to handle it, so have not listed them here, but obviously would be happy to do so.
Will, the driver author, is on cc and would be happy to give advice regarding dependencies, etc. But he has other work and so can’t take a big support burden here.
I’m assuming this needs to go to TSC for approval – so I’ve cc’ed Roger and Mark.
Thanks, James
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
On Mon, Jan 19, 2015 at 04:38:04PM +0000, James King wrote:
Any Juno board should be fine as that has a SMMUv1. Once in LSK, we can test it is still ok on a v2 part.
Will – are there some tests (LTP?) that can be used to test it is working ok when back ported?
Nope; there are no in-tree users of the ARM SMMU, mainly due to lack of IOMMU-backed DMA ops for arm64 (patches are in review stage atm).
You could have a play with VFIO, but you'll need userspace support and working PCI for that.
Will
Hi Mark,
In which case, the best we can request now is a back-port that builds.
I will be giving the kernel to developers of a display driver that want to use it and can test it out and provide any fixes required.
Thanks, James
-----Original Message----- From: Will Deacon [mailto:will.deacon@arm.com] Sent: 19 January 2015 16:48 To: James King Cc: Mark Brown; alex.shi@linaro.org; linaro-kernel@lists.linaro.org; Mark Hambleton; Roger Teague Subject: Re: 3.10 LSK request
On Mon, Jan 19, 2015 at 04:38:04PM +0000, James King wrote:
Any Juno board should be fine as that has a SMMUv1. Once in LSK, we can test it is still ok on a v2 part.
Will – are there some tests (LTP?) that can be used to test it is working ok when back ported?
Nope; there are no in-tree users of the ARM SMMU, mainly due to lack of IOMMU-backed DMA ops for arm64 (patches are in review stage atm).
You could have a play with VFIO, but you'll need userspace support and working PCI for that.
Will
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
Hi James,
Can we at least get the DT information required to load the driver (ideally in mainline so we can backport it and we get any upstream testing going)? It's obviously not terribly meaningful as testing goes but it's low effort and better than nothing.
Thanks, Mark
On 19 January 2015 at 16:51, James King James.King@arm.com wrote:
Hi Mark,
In which case, the best we can request now is a back-port that builds.
I will be giving the kernel to developers of a display driver that want to use it and can test it out and provide any fixes required.
Thanks, James
-----Original Message----- From: Will Deacon [mailto:will.deacon@arm.com] Sent: 19 January 2015 16:48 To: James King Cc: Mark Brown; alex.shi@linaro.org; linaro-kernel@lists.linaro.org; Mark Hambleton; Roger Teague Subject: Re: 3.10 LSK request
On Mon, Jan 19, 2015 at 04:38:04PM +0000, James King wrote:
Any Juno board should be fine as that has a SMMUv1. Once in LSK, we can test it is still ok on a v2 part.
Will – are there some tests (LTP?) that can be used to test it is working ok when back ported?
Nope; there are no in-tree users of the ARM SMMU, mainly due to lack of IOMMU-backed DMA ops for arm64 (patches are in review stage atm).
You could have a play with VFIO, but you'll need userspace support and working PCI for that.
Will
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
On Mon, Jan 19, 2015 at 04:59:25PM +0000, Mark Brown wrote:
Can we at least get the DT information required to load the driver (ideally in mainline so we can backport it and we get any upstream testing going)? It's obviously not terribly meaningful as testing goes but it's low effort and better than nothing.
The other snag there is that there is discussion upstream about changing the binding entirely (to use the generic IOMMU bindings from Thierry). This has been deemed OK as there are no users of the current binding, so if Juno is going to become the first user then we'd need to discuss upstream.
Will
On 19 January 2015 at 17:03, Will Deacon will.deacon@arm.com wrote:
On Mon, Jan 19, 2015 at 04:59:25PM +0000, Mark Brown wrote:
Can we at least get the DT information required to load the driver (ideally in mainline so we can backport it and we get any upstream
testing
going)? It's obviously not terribly meaningful as testing goes but it's low effort and better than nothing.
The other snag there is that there is discussion upstream about changing the binding entirely (to use the generic IOMMU bindings from Thierry). This has been deemed OK as there are no users of the current binding, so if Juno is going to become the first user then we'd need to discuss upstream.
I'd prefer to have at least some of that discussion before backporting - if people start using LSK DTs we may end up having to support the current binding in mainline anyway so it'd seem better to backport with the generic binding if we can.
linaro-kernel@lists.linaro.org
-
James King -
Mark Brown -
Will Deacon