i915 drm patches in LTS 4.4.x ?

i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

I notice there are no I915 DRM patches in going into
4.4.x for a while. I wonder if the inflow of such patches
has stopped or if they are just batched/saved for later?

ï¿œJocke

Re: i915 drm patches in LTS 4.4.x ?

[Greg KH]

On Mon, Jul 18, 2016 at 02:21:45PM +0000, Joakim Tjernlund wrote:
> I notice there are no I915 DRM patches in going into
> 4.4.x for a while.

Define "a while".

> I wonder if the inflow of such patches has stopped or if they are just
> batched/saved for later?

I only see 5 pending i915 patches in my stable queue at the moment, not
many at all.  Was there something specific you were looking for that did
not get merged?

thanks,

greg k-h

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

On Mon, 2016-07-18 at 13:52 -0700, Greg KH wrote:
> On Mon, Jul 18, 2016 at 02:21:45PM +0000, Joakim Tjernlund wrote:
> > 
> > I notice there are no I915 DRM patches in going into
> > 4.4.x for a while.
> 
> Define "a while".

Not since 4.4.13, may I missed some path though.

> 
> > 
> > I wonder if the inflow of such patches has stopped or if they are just
> > batched/saved for later?
> 
> I only see 5 pending i915 patches in my stable queue at the moment, not

I have a hard time seeing any i915 patches in your queue, must get glasses :)

> many at all.ￜￜWas there something specific you were looking for that did
> not get merged?
>ￜ

We have some HP Broadwell laptops which has a problem with external monitors,
they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
One path caught my eye though was this:
ￜ drm/i915: Revert DisplayPort fast link training feature
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511dce4b35685d3988d5c8b100d11a068db5bd

Is that one in your queue?

Re: i915 drm patches in LTS 4.4.x ?

[Willy Tarreau]

On Mon, Jul 18, 2016 at 11:41:02PM +0000, Joakim Tjernlund wrote:
> On Mon, 2016-07-18 at 13:52 -0700, Greg KH wrote:
> > On Mon, Jul 18, 2016 at 02:21:45PM +0000, Joakim Tjernlund wrote:
> > > I wonder if the inflow of such patches has stopped or if they are just
> > > batched/saved for later?
> > 
> > I only see 5 pending i915 patches in my stable queue at the moment, not
> 
> I have a hard time seeing any i915 patches in your queue, must get glasses :)

Those you see are the ones Greg could already merge in his branches. He
sees them in his mailbox.

> > many at all.��Was there something specific you were looking for that did
> > not get merged?
> >�
> 
> We have some HP Broadwell laptops which has a problem with external monitors,
> they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> One path caught my eye though was this:
> � drm/i915: Revert DisplayPort fast link training feature
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511dce4b35685d3988d5c8b100d11a068db5bd

Did you try it or at least the kernel which contains it ? It is important
for stable branches (especially LTS ones) to have a good reason for
backporting a patch because there's always a risk to break some setups.
If you know the patch above improves your situation, it's a lot better
than just hoping it will improve it.

Overall the process is simple : if the latest kernel still doesn't work
for you, you want to report your breakage to the subsystem maintainers
who may have an idea how to fix it and who may request further info from
you. If the latest kernel works, you can try to find which patch made it
work (git bisect helps here) and ask for its backport. If finding which
patch fixes your problem is too difficult, asking the subsystem maintainers
can often help because if they recently fixed the problem they'll often
recognize the same issue in your description and will spot the relevant
patch.

Regards,
Willy

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

On Tue, 2016-07-19 at 07:26 +0200, Willy Tarreau wrote:
> On Mon, Jul 18, 2016 at 11:41:02PM +0000, Joakim Tjernlund wrote:
> > 
> > On Mon, 2016-07-18 at 13:52 -0700, Greg KH wrote:
> > > 
> > > On Mon, Jul 18, 2016 at 02:21:45PM +0000, Joakim Tjernlund wrote:
> > > > 
> > > > I wonder if the inflow of such patches has stopped or if they are just
> > > > batched/saved for later?
> > > 
> > > I only see 5 pending i915 patches in my stable queue at the moment, not
> > 
> > I have a hard time seeing any i915 patches in your queue, must get glasses :)
> 
> Those you see are the ones Greg could already merge in his branches. He
> sees them in his mailbox.

I see

> 
> > 
> > > 
> > > many at all.ￜￜWas there something specific you were looking for that did
> > > not get merged?
> > > ￜ
> > 
> > We have some HP Broadwell laptops which has a problem with external monitors,
> > they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> > One path caught my eye though was this:
> > ￜ drm/i915: Revert DisplayPort fast link training feature
> > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511dce4b3
> > 5685d3988d5c8b100d11a068db5bd
> 
> Did you try it or at least the kernel which contains it ? It is important

No I haven't. I myself do not have such a laptop so I must find someone that does
and willing to test stuff for me(I maintain our Linux DE used for in-house development)
All I can say for sure is that 4.1.x works and 4.4.x does not :(

> for stable branches (especially LTS ones) to have a good reason for
> backporting a patch because there's always a risk to break some setups.
> If you know the patch above improves your situation, it's a lot better
> than just hoping it will improve it.
> 
> Overall the process is simple : if the latest kernel still doesn't work
> for you, you want to report your breakage to the subsystem maintainers
> who may have an idea how to fix it and who may request further info from
> you. If the latest kernel works, you can try to find which patch made it
> work (git bisect helps here) and ask for its backport. If finding which
> patch fixes your problem is too difficult, asking the subsystem maintainers
> can often help because if they recently fixed the problem they'll often
> recognize the same issue in your description and will spot the relevant
> patch.
>ￜ

Yes I get that but I figured that a upstream commit that is marked:
ï¿œCc: <stable@vger.kernel.org> # v4.2+
already was approved for stable and it would find its way into all relevant branches.
But perhaps I am wrong and you need further testing in each relevant branch?

ￜ ￜJocke

Re: i915 drm patches in LTS 4.4.x ?

[Willy Tarreau]

On Tue, Jul 19, 2016 at 08:41:21AM +0000, Joakim Tjernlund wrote:
> > Did you try it or at least the kernel which contains it ? It is important
> 
> No I haven't. I myself do not have such a laptop so I must find someone that does
> and willing to test stuff for me(I maintain our Linux DE used for in-house development)
> All I can say for sure is that 4.1.x works and 4.4.x does not :(

OK that's already a good indication.

(...)
> Yes I get that but I figured that a upstream commit that is marked:
> �Cc: <stable@vger.kernel.org> # v4.2+
> already was approved for stable and it would find its way into all relevant branches.

Indeed it could be one candidate, so if you can afford to wait for a
next release to see if it fixes your problem that's the easiest way.
Otherwise you'll have to bisect either between 4.1 and 4.4 to find
which one broke, or to test mainline to see if it's fixed again. But
given that it was a regression, it's likely that someone else will
notice it and propose a fix (possibly the one above).

Willy

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

On Tue, 2016-07-19 at 10:59 +0200, Willy Tarreau wrote:
> On Tue, Jul 19, 2016 at 08:41:21AM +0000, Joakim Tjernlund wrote:
> > 
> > > 
> > > Did you try it or at least the kernel which contains it ? It is important
> > 
> > No I haven't. I myself do not have such a laptop so I must find someone that does
> > and willing to test stuff for me(I maintain our Linux DE used for in-house development)
> > All I can say for sure is that 4.1.x works and 4.4.x does not :(
> 
> OK that's already a good indication.
> 
> (...)
> > 
> > Yes I get that but I figured that a upstream commit that is marked:
> > ï¿œCc: <stable@vger.kernel.org> # v4.2+
> > already was approved for stable and it would find its way into all relevant branches.
> 
> Indeed it could be one candidate, so if you can afford to wait for a
> next release to see if it fixes your problem that's the easiest way.
> Otherwise you'll have to bisect either between 4.1 and 4.4 to find
> which one broke, or to test mainline to see if it's fixed again. But
> given that it was a regression, it's likely that someone else will
> notice it and propose a fix (possibly the one above).
>ￜ

Yes, plenty of people have similar has a problem but Intel devs have had hard time
finding the bugs.

Greg, could push your mail queue so I can compare with upstream?

Seems like this patch has been lost though:
ￜ[PATCH stable-4.4] drm/i915: Update CDCLK_FREQ register on BDW after changing cdclk frequency

ï¿œJocke

Re: i915 drm patches in LTS 4.4.x ?

[gregkh]

On Mon, Jul 18, 2016 at 11:41:02PM +0000, Joakim Tjernlund wrote:
> On Mon, 2016-07-18 at 13:52 -0700, Greg KH wrote:
> > On Mon, Jul 18, 2016 at 02:21:45PM +0000, Joakim Tjernlund wrote:
> > > 
> > > I notice there are no I915 DRM patches in going into
> > > 4.4.x for a while.
> > 
> > Define "a while".
> 
> Not since 4.4.13, may I missed some path though.
> 
> > 
> > > 
> > > I wonder if the inflow of such patches has stopped or if they are just
> > > batched/saved for later?
> > 
> > I only see 5 pending i915 patches in my stable queue at the moment, not
> 
> I have a hard time seeing any i915 patches in your queue, must get glasses :)
> 
> > many at all.��Was there something specific you were looking for that did
> > not get merged?
> >�
> 
> We have some HP Broadwell laptops which has a problem with external monitors,
> they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> One path caught my eye though was this:
> � drm/i915: Revert DisplayPort fast link training feature
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511dce4b35685d3988d5c8b100d11a068db5bd
> 
> Is that one in your queue?

Yes it is, I have not gotten to that one yet.

thanks,

greg k-h

Re: i915 drm patches in LTS 4.4.x ?

[gregkh]

On Tue, Jul 19, 2016 at 04:43:06PM +0000, Joakim Tjernlund wrote:
> On Tue, 2016-07-19 at 10:59 +0200, Willy Tarreau wrote:
> > On Tue, Jul 19, 2016 at 08:41:21AM +0000, Joakim Tjernlund wrote:
> > > 
> > > > 
> > > > Did you try it or at least the kernel which contains it ? It is important
> > > 
> > > No I haven't. I myself do not have such a laptop so I must find someone that does
> > > and willing to test stuff for me(I maintain our Linux DE used for in-house development)
> > > All I can say for sure is that 4.1.x works and 4.4.x does not :(
> > 
> > OK that's already a good indication.
> > 
> > (...)
> > > 
> > > Yes I get that but I figured that a upstream commit that is marked:
> > > �Cc: <stable@vger.kernel.org> # v4.2+
> > > already was approved for stable and it would find its way into all relevant branches.
> > 
> > Indeed it could be one candidate, so if you can afford to wait for a
> > next release to see if it fixes your problem that's the easiest way.
> > Otherwise you'll have to bisect either between 4.1 and 4.4 to find
> > which one broke, or to test mainline to see if it's fixed again. But
> > given that it was a regression, it's likely that someone else will
> > notice it and propose a fix (possibly the one above).
> >�
> 
> Yes, plenty of people have similar has a problem but Intel devs have had hard time
> finding the bugs.
> 
> Greg, could push your mail queue so I can compare with upstream?

I don't know what you mean by that, I'm not going to push my stable mbox
up publicly, sorry :)

> Seems like this patch has been lost though:
> �[PATCH stable-4.4] drm/i915: Update CDCLK_FREQ register on BDW after changing cdclk frequency

Not lost, still in my todo queue, along with 663 other patches I need to
review/apply...

thanks,

greg k-h

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

On Tue, 2016-07-19 at 15:06 -0700, gregkh@linuxfoundation.org wrote:
> On Mon, Jul 18, 2016 at 11:41:02PM +0000, Joakim Tjernlund wrote:
> > 
> > On Mon, 2016-07-18 at 13:52 -0700, Greg KH wrote:
> > > 
> > > On Mon, Jul 18, 2016 at 02:21:45PM +0000, Joakim Tjernlund wrote:
> > > > 
> > > > 
> > > > I notice there are no I915 DRM patches in going into
> > > > 4.4.x for a while.
> > > 
> > > Define "a while".
> > 
> > Not since 4.4.13, may I missed some path though.
> > 
> > > 
> > > 
> > > > 
> > > > 
> > > > I wonder if the inflow of such patches has stopped or if they are just
> > > > batched/saved for later?
> > > 
> > > I only see 5 pending i915 patches in my stable queue at the moment, not
> > 
> > I have a hard time seeing any i915 patches in your queue, must get glasses :)
> > 
> > > 
> > > many at all.ￜￜWas there something specific you were looking for that did
> > > not get merged?
> > > ￜ
> > 
> > We have some HP Broadwell laptops which has a problem with external monitors,
> > they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> > One path caught my eye though was this:
> > ￜ drm/i915: Revert DisplayPort fast link training feature
> > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511dce4b3
> > 5685d3988d5c8b100d11a068db5bd
> > 
> > Is that one in your queue?
> 
> Yes it is, I have not gotten to that one yet.
>ￜ

I did a quick backport of above patch for 4.4 and had a guy test it his laptop and it did
fix the problem :)

ￜ ￜ Jocke

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

On Tue, 2016-07-19 at 15:08 -0700, gregkh@linuxfoundation.org wrote:
> On Tue, Jul 19, 2016 at 04:43:06PM +0000, Joakim Tjernlund wrote:
> > 
> > On Tue, 2016-07-19 at 10:59 +0200, Willy Tarreau wrote:
> > > 
> > > On Tue, Jul 19, 2016 at 08:41:21AM +0000, Joakim Tjernlund wrote:
> > > > 
> > > > 
> > > > > 
> > > > > 
> > > > > Did you try it or at least the kernel which contains it ? It is important
> > > > 
> > > > No I haven't. I myself do not have such a laptop so I must find someone that does
> > > > and willing to test stuff for me(I maintain our Linux DE used for in-house development)
> > > > All I can say for sure is that 4.1.x works and 4.4.x does not :(
> > > 
> > > OK that's already a good indication.
> > > 
> > > (...)
> > > > 
> > > > 
> > > > Yes I get that but I figured that a upstream commit that is marked:
> > > > ï¿œCc: <stable@vger.kernel.org> # v4.2+
> > > > already was approved for stable and it would find its way into all relevant branches.
> > > 
> > > Indeed it could be one candidate, so if you can afford to wait for a
> > > next release to see if it fixes your problem that's the easiest way.
> > > Otherwise you'll have to bisect either between 4.1 and 4.4 to find
> > > which one broke, or to test mainline to see if it's fixed again. But
> > > given that it was a regression, it's likely that someone else will
> > > notice it and propose a fix (possibly the one above).
> > > ￜ
> > 
> > Yes, plenty of people have similar has a problem but Intel devs have had hard time
> > finding the bugs.
> > 
> > Greg, could push your mail queue so I can compare with upstream?
> 
> I don't know what you mean by that, I'm not going to push my stable mbox
> up publicly, sorry :)

hehe, not your mbox :) You had 5 pending i915 drm patches in your queue you said earlier.
Please push these toï¿œhttps://git.kernel.org/cgit/linux/kernel/git/stable/stable-queue.git/log/
especially the 4.4 branch :)

> 
> > 
> > Seems like this patch has been lost though:
> > ￜ[PATCH stable-4.4] drm/i915: Update CDCLK_FREQ register on BDW after changing cdclk frequency
> 
> Not lost, still in my todo queue, along with 663 other patches I need to
> review/apply...

Ouch, that is a lot of patches. Does the queue ever get smaller?

ï¿œJocke

Re: i915 drm patches in LTS 4.4.x ?

[Willy Tarreau]

On Wed, Jul 20, 2016 at 08:49:36AM +0000, Joakim Tjernlund wrote:
> On Tue, 2016-07-19 at 15:08 -0700, gregkh@linuxfoundation.org wrote:
> > On Tue, Jul 19, 2016 at 04:43:06PM +0000, Joakim Tjernlund wrote:
> > > Greg, could push your mail queue so I can compare with upstream?
> > 
> > I don't know what you mean by that, I'm not going to push my stable mbox
> > up publicly, sorry :)
> 
> hehe, not your mbox :) You had 5 pending i915 drm patches in your queue you said earlier.
> Please push these to�https://git.kernel.org/cgit/linux/kernel/git/stable/stable-queue.git/log/
> especially the 4.4 branch :)

It seems to me you don't realize that there is a substantial difference
between seeing patches in one's mailbox and applying them to a tree
fixing the conflicts to then be able to add them to a queue. For the
same reason I can say I have a few patches pending for 3.10 but they're
not in a queue yet and will only be once I process them.

Willy

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

On Wed, 2016-07-20 at 10:58 +0200, Willy Tarreau wrote:
> On Wed, Jul 20, 2016 at 08:49:36AM +0000, Joakim Tjernlund wrote:
> > 
> > On Tue, 2016-07-19 at 15:08 -0700, gregkh@linuxfoundation.org wrote:
> > > 
> > > On Tue, Jul 19, 2016 at 04:43:06PM +0000, Joakim Tjernlund wrote:
> > > > 
> > > > Greg, could push your mail queue so I can compare with upstream?
> > > 
> > > I don't know what you mean by that, I'm not going to push my stable mbox
> > > up publicly, sorry :)
> > 
> > hehe, not your mbox :) You had 5 pending i915 drm patches in your queue you said earlier.
> > Please push these toï¿œhttps://git.kernel.org/cgit/linux/kernel/git/stable/stable-queue.git/log/
> > especially the 4.4 branch :)
> 
> It seems to me you don't realize that there is a substantial difference
> between seeing patches in one's mailbox and applying them to a tree
> fixing the conflicts to then be able to add them to a queue. For the
> same reason I can say I have a few patches pending for 3.10 but they're
> not in a queue yet and will only be once I process them.
>ￜ

I do but I can be a bit too direct and I am a lousy typewriter ...
Let me rephrase a bit:
Please consider moving those 5 i915 drm patches to the top of the queue
as at least one of them fixes a long standing regression which prevents our
Broadwell laptop users to move to 4.4

ï¿œJockeï¿œ

Re: i915 drm patches in LTS 4.4.x ?

[gregkh]

On Wed, Jul 20, 2016 at 08:41:06AM +0000, Joakim Tjernlund wrote:
> On Tue, 2016-07-19 at 15:06 -0700, gregkh@linuxfoundation.org wrote:
> > On Mon, Jul 18, 2016 at 11:41:02PM +0000, Joakim Tjernlund wrote:
> > > 
> > > On Mon, 2016-07-18 at 13:52 -0700, Greg KH wrote:
> > > > 
> > > > On Mon, Jul 18, 2016 at 02:21:45PM +0000, Joakim Tjernlund wrote:
> > > > > 
> > > > > 
> > > > > I notice there are no I915 DRM patches in going into
> > > > > 4.4.x for a while.
> > > > 
> > > > Define "a while".
> > > 
> > > Not since 4.4.13, may I missed some path though.
> > > 
> > > > 
> > > > 
> > > > > 
> > > > > 
> > > > > I wonder if the inflow of such patches has stopped or if they are just
> > > > > batched/saved for later?
> > > > 
> > > > I only see 5 pending i915 patches in my stable queue at the moment, not
> > > 
> > > I have a hard time seeing any i915 patches in your queue, must get glasses :)
> > > 
> > > > 
> > > > many at all.��Was there something specific you were looking for that did
> > > > not get merged?
> > > > �
> > > 
> > > We have some HP Broadwell laptops which has a problem with external monitors,
> > > they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> > > One path caught my eye though was this:
> > > � drm/i915: Revert DisplayPort fast link training feature
> > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511dce4b3
> > > 5685d3988d5c8b100d11a068db5bd
> > > 
> > > Is that one in your queue?
> > 
> > Yes it is, I have not gotten to that one yet.
> >�
> 
> I did a quick backport of above patch for 4.4 and had a guy test it his laptop and it did
> fix the problem :)

Can you provide the backport for 4.4 and 4.6-stable trees, as it
currently does not apply for me at all :(

thanks,

greg k-h

Re: i915 drm patches in LTS 4.4.x ?

[gregkh]

On Wed, Jul 20, 2016 at 10:48:34AM +0000, Joakim Tjernlund wrote:
> On Wed, 2016-07-20 at 10:58 +0200, Willy Tarreau wrote:
> > On Wed, Jul 20, 2016 at 08:49:36AM +0000, Joakim Tjernlund wrote:
> > > 
> > > On Tue, 2016-07-19 at 15:08 -0700, gregkh@linuxfoundation.org wrote:
> > > > 
> > > > On Tue, Jul 19, 2016 at 04:43:06PM +0000, Joakim Tjernlund wrote:
> > > > > 
> > > > > Greg, could push your mail queue so I can compare with upstream?
> > > > 
> > > > I don't know what you mean by that, I'm not going to push my stable mbox
> > > > up publicly, sorry :)
> > > 
> > > hehe, not your mbox :) You had 5 pending i915 drm patches in your queue you said earlier.
> > > Please push these to�https://git.kernel.org/cgit/linux/kernel/git/stable/stable-queue.git/log/
> > > especially the 4.4 branch :)
> > 
> > It seems to me you don't realize that there is a substantial difference
> > between seeing patches in one's mailbox and applying them to a tree
> > fixing the conflicts to then be able to add them to a queue. For the
> > same reason I can say I have a few patches pending for 3.10 but they're
> > not in a queue yet and will only be once I process them.
> >�
> 
> I do but I can be a bit too direct and I am a lousy typewriter ...
> Let me rephrase a bit:
> Please consider moving those 5 i915 drm patches to the top of the queue
> as at least one of them fixes a long standing regression which prevents our
> Broadwell laptop users to move to 4.4

That last patch doesn't even apply, so I can't merge it now, a backport
would be appreciated.

Also, picking which ones go on the top of the queue is hard, I usually
have to take them all in order so that they apply correctly, but I've
now caught up on all drm patches, so if there is anything missing that
you see, please let me know.

thanks,

greg k-h

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

[-- Attachment #1: Type: text/plain, Size: 1235 bytes --]

On Sun, 2016-07-24 at 16:41 -0700, gregkh@linuxfoundation.org wrote:
> On Wed, Jul 20, 2016 at 08:41:06AM +0000, Joakim Tjernlund wrote:
> > 
> > On Tue, 2016-07-19 at 15:06 -0700, gregkh@linuxfoundation.org wrote:
....

> > >  which has a problem with external monitors,
> > > > they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> > > > One path caught my eye though was this:
> > > >   drm/i915: Revert DisplayPort fast link training feature
> > > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511dc
> > > > e4b3
> > > > 5685d3988d5c8b100d11a068db5bd
> > > > 
> > > > Is that one in your queue?
> > > 
> > > Yes it is, I have not gotten to that one yet.
> > >  
> > 
> > I did a quick backport of above patch for 4.4 and had a guy test it his laptop and it did
> > fix the problem :)
> 
> Can you provide the backport for 4.4 and 4.6-stable trees, as it
> currently does not apply for me at all :(
> 

Just did a backport to 4.4( I hope it applies to 4.6 as well:) but git send email does not work
ATM because we are moving to a new mail domain and it is not finished yet. Attaching the patch instead.
 

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-drm-i915-Revert-DisplayPort-fast-link-training-featu.patch --]
[-- Type: text/x-patch; name="0001-drm-i915-Revert-DisplayPort-fast-link-training-featu.patch", Size: 4568 bytes --]

From c403c1a872705d440a4c03b14618adfb36c93568 Mon Sep 17 00:00:00 2001
From: Mika Kahola <mika.kahola@intel.com>
Date: Mon, 20 Jun 2016 11:10:26 +0300
Subject: [PATCH] drm/i915: Revert DisplayPort fast link training feature

It has been found out that in some HW combination the DisplayPort
fast link training feature caused screen flickering. Let's revert
this feature for now until we can ensure that the feature works for
all platforms.

This is a manual revert of commits 5fa836a9d859 ("drm/i915: DP link
training optimization") and 4e96c97742f4 ("drm/i915: eDP link training
optimization").

Fixes: 5fa836a9d859 ("drm/i915: DP link training optimization")
Fixes: 4e96c97742f4 ("drm/i915: eDP link training optimization")
Cc: <stable@vger.kernel.org>

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91393
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1466410226-19543-1-git-send-email-mika.kahola@intel.com
(cherry picked from commit 91df09d92ad82c8778ca218097bf827f154292ca)
Signed-off-by: Joakim Tjernlund <joakim.tjernlund@infinera.com>
---

 Backported to 4.4.y branch

 drivers/gpu/drm/i915/intel_dp.c  | 28 ++--------------------------
 drivers/gpu/drm/i915/intel_drv.h |  1 -
 2 files changed, 2 insertions(+), 27 deletions(-)

diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index e55a82a..dd5cfd2 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -3628,8 +3628,7 @@ static bool
 intel_dp_reset_link_train(struct intel_dp *intel_dp, uint32_t *DP,
 			uint8_t dp_train_pat)
 {
-	if (!intel_dp->train_set_valid)
-		memset(intel_dp->train_set, 0, sizeof(intel_dp->train_set));
+	memset(intel_dp->train_set, 0, sizeof(intel_dp->train_set));
 	intel_dp_set_signal_levels(intel_dp, DP);
 	return intel_dp_set_link_train(intel_dp, DP, dp_train_pat);
 }
@@ -3746,22 +3745,6 @@ intel_dp_link_training_clock_recovery(struct intel_dp *intel_dp)
 			break;
 		}
 
-		/*
-		 * if we used previously trained voltage and pre-emphasis values
-		 * and we don't get clock recovery, reset link training values
-		 */
-		if (intel_dp->train_set_valid) {
-			DRM_DEBUG_KMS("clock recovery not ok, reset");
-			/* clear the flag as we are not reusing train set */
-			intel_dp->train_set_valid = false;
-			if (!intel_dp_reset_link_train(intel_dp, &DP,
-						       DP_TRAINING_PATTERN_1 |
-						       DP_LINK_SCRAMBLING_DISABLE)) {
-				DRM_ERROR("failed to enable link training\n");
-				return;
-			}
-			continue;
-		}
 
 		/* Check to see if we've tried the max voltage */
 		for (i = 0; i < intel_dp->lane_count; i++)
@@ -3854,7 +3837,6 @@ intel_dp_link_training_channel_equalization(struct intel_dp *intel_dp)
 		/* Make sure clock is still ok */
 		if (!drm_dp_clock_recovery_ok(link_status,
 					      intel_dp->lane_count)) {
-			intel_dp->train_set_valid = false;
 			intel_dp_link_training_clock_recovery(intel_dp);
 			intel_dp_set_link_train(intel_dp, &DP,
 						training_pattern |
@@ -3871,7 +3853,6 @@ intel_dp_link_training_channel_equalization(struct intel_dp *intel_dp)
 
 		/* Try 5 times, then try clock recovery if that fails */
 		if (tries > 5) {
-			intel_dp->train_set_valid = false;
 			intel_dp_link_training_clock_recovery(intel_dp);
 			intel_dp_set_link_train(intel_dp, &DP,
 						training_pattern |
@@ -3893,10 +3874,8 @@ intel_dp_link_training_channel_equalization(struct intel_dp *intel_dp)
 
 	intel_dp->DP = DP;
 
-	if (channel_eq) {
-		intel_dp->train_set_valid = true;
+	if (channel_eq)
 		DRM_DEBUG_KMS("Channel EQ done. DP Training successful\n");
-	}
 }
 
 void intel_dp_stop_link_train(struct intel_dp *intel_dp)
@@ -5157,9 +5136,6 @@ intel_dp_hpd_pulse(struct intel_digital_port *intel_dig_port, bool long_hpd)
 	intel_display_power_get(dev_priv, power_domain);
 
 	if (long_hpd) {
-		/* indicate that we need to restart link training */
-		intel_dp->train_set_valid = false;
-
 		if (!intel_digital_port_connected(dev_priv, intel_dig_port))
 			goto mst_fail;
 
diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
index f34a219..c5f11e0 100644
--- a/drivers/gpu/drm/i915/intel_drv.h
+++ b/drivers/gpu/drm/i915/intel_drv.h
@@ -783,7 +783,6 @@ struct intel_dp {
 				     bool has_aux_irq,
 				     int send_bytes,
 				     uint32_t aux_clock_divider);
-	bool train_set_valid;
 
 	/* Displayport compliance testing */
 	unsigned long compliance_test_type;
-- 
2.7.3

Re: i915 drm patches in LTS 4.4.x ?

[Joakim Tjernlund]

[-- Attachment #1: Type: text/plain, Size: 1468 bytes --]

On Mon, 2016-07-25 at 13:39 +0200, Joakim Tjernlund wrote:
> On Sun, 2016-07-24 at 16:41 -0700, gregkh@linuxfoundation.org wrote:
> > 
> > On Wed, Jul 20, 2016 at 08:41:06AM +0000, Joakim Tjernlund wrote:
> > > 
> > > 
> > > On Tue, 2016-07-19 at 15:06 -0700, gregkh@linuxfoundation.org wrote:
> ....
> 
> > 
> > > 
> > > > 
> > > >  which has a problem with external monitors,
> > > > > 
> > > > > they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> > > > > One path caught my eye though was this:
> > > > >   drm/i915: Revert DisplayPort fast link training feature
> > > > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511
> > > > > dc
> > > > > e4b3
> > > > > 5685d3988d5c8b100d11a068db5bd
> > > > > 
> > > > > Is that one in your queue?
> > > > 
> > > > Yes it is, I have not gotten to that one yet.
> > > >  
> > > 
> > > I did a quick backport of above patch for 4.4 and had a guy test it his laptop and it did
> > > fix the problem :)
> > 
> > Can you provide the backport for 4.4 and 4.6-stable trees, as it
> > currently does not apply for me at all :(
> >  
> 
> Just did a backport to 4.4( I hope it applies to 4.6 as well:) but git send email does not work
> ATM because we are moving to a new mail domain and it is not finished yet. Attaching the patch instead.

Well, it did not apply to 4.6.y so here is backport

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-4.6.y-drm-i915-Revert-DisplayPort-fast-link-training.patch --]
[-- Type: text/x-patch; name="0001-4.6.y-drm-i915-Revert-DisplayPort-fast-link-training.patch", Size: 4917 bytes --]

From 3cb21630b6530687c6b4f535811e7a24ec0596ae Mon Sep 17 00:00:00 2001
From: Joakim Tjernlund <joakim.tjernlund@infinera.com>
Date: Mon, 25 Jul 2016 13:55:06 +0200
Subject: [PATCH] [4.6.y] drm/i915: Revert DisplayPort fast link training
 feature

It has been found out that in some HW combination the DisplayPort
fast link training feature caused screen flickering. Let's revert
this feature for now until we can ensure that the feature works for
all platforms.

This is a manual revert of commits 5fa836a9d859 ("drm/i915: DP link
training optimization") and 4e96c97742f4 ("drm/i915: eDP link training
optimization").

Fixes: 5fa836a9d859 ("drm/i915: DP link training optimization")
Fixes: 4e96c97742f4 ("drm/i915: eDP link training optimization")
Cc: <stable@vger.kernel.org>

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91393
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1466410226-19543-1-git-send-email-mika.kahola@intel.com
(cherry picked from commit 91df09d92ad82c8778ca218097bf827f154292ca)
Signed-off-by: Joakim Tjernlund <joakim.tjernlund@infinera.com>
---
 drivers/gpu/drm/i915/intel_dp.c               |  3 ---
 drivers/gpu/drm/i915/intel_dp_link_training.c | 25 ++-----------------------
 drivers/gpu/drm/i915/intel_drv.h              |  2 --
 3 files changed, 2 insertions(+), 28 deletions(-)

diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index 412a34c..8e1f186 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -5020,9 +5020,6 @@ intel_dp_hpd_pulse(struct intel_digital_port *intel_dig_port, bool long_hpd)
 	intel_display_power_get(dev_priv, power_domain);
 
 	if (long_hpd) {
-		/* indicate that we need to restart link training */
-		intel_dp->train_set_valid = false;
-
 		if (!intel_digital_port_connected(dev_priv, intel_dig_port))
 			goto mst_fail;
 
diff --git a/drivers/gpu/drm/i915/intel_dp_link_training.c b/drivers/gpu/drm/i915/intel_dp_link_training.c
index 0b8eefc..926a1e6 100644
--- a/drivers/gpu/drm/i915/intel_dp_link_training.c
+++ b/drivers/gpu/drm/i915/intel_dp_link_training.c
@@ -85,8 +85,7 @@ static bool
 intel_dp_reset_link_train(struct intel_dp *intel_dp,
 			uint8_t dp_train_pat)
 {
-	if (!intel_dp->train_set_valid)
-		memset(intel_dp->train_set, 0, sizeof(intel_dp->train_set));
+	memset(intel_dp->train_set, 0, sizeof(intel_dp->train_set));
 	intel_dp_set_signal_levels(intel_dp);
 	return intel_dp_set_link_train(intel_dp, dp_train_pat);
 }
@@ -161,22 +160,6 @@ intel_dp_link_training_clock_recovery(struct intel_dp *intel_dp)
 			break;
 		}
 
-		/*
-		 * if we used previously trained voltage and pre-emphasis values
-		 * and we don't get clock recovery, reset link training values
-		 */
-		if (intel_dp->train_set_valid) {
-			DRM_DEBUG_KMS("clock recovery not ok, reset");
-			/* clear the flag as we are not reusing train set */
-			intel_dp->train_set_valid = false;
-			if (!intel_dp_reset_link_train(intel_dp,
-						       DP_TRAINING_PATTERN_1 |
-						       DP_LINK_SCRAMBLING_DISABLE)) {
-				DRM_ERROR("failed to enable link training\n");
-				return;
-			}
-			continue;
-		}
 
 		/* Check to see if we've tried the max voltage */
 		for (i = 0; i < intel_dp->lane_count; i++)
@@ -284,7 +267,6 @@ intel_dp_link_training_channel_equalization(struct intel_dp *intel_dp)
 		/* Make sure clock is still ok */
 		if (!drm_dp_clock_recovery_ok(link_status,
 					      intel_dp->lane_count)) {
-			intel_dp->train_set_valid = false;
 			intel_dp_link_training_clock_recovery(intel_dp);
 			intel_dp_set_link_train(intel_dp,
 						training_pattern |
@@ -301,7 +283,6 @@ intel_dp_link_training_channel_equalization(struct intel_dp *intel_dp)
 
 		/* Try 5 times, then try clock recovery if that fails */
 		if (tries > 5) {
-			intel_dp->train_set_valid = false;
 			intel_dp_link_training_clock_recovery(intel_dp);
 			intel_dp_set_link_train(intel_dp,
 						training_pattern |
@@ -322,10 +303,8 @@ intel_dp_link_training_channel_equalization(struct intel_dp *intel_dp)
 
 	intel_dp_set_idle_link_train(intel_dp);
 
-	if (channel_eq) {
-		intel_dp->train_set_valid = true;
+	if (channel_eq)
 		DRM_DEBUG_KMS("Channel EQ done. DP Training successful\n");
-	}
 }
 
 void intel_dp_stop_link_train(struct intel_dp *intel_dp)
diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
index 3a30b37..8dd2cc5 100644
--- a/drivers/gpu/drm/i915/intel_drv.h
+++ b/drivers/gpu/drm/i915/intel_drv.h
@@ -811,8 +811,6 @@ struct intel_dp {
 	/* This is called before a link training is starterd */
 	void (*prepare_link_retrain)(struct intel_dp *intel_dp);
 
-	bool train_set_valid;
-
 	/* Displayport compliance testing */
 	unsigned long compliance_test_type;
 	unsigned long compliance_test_data;
-- 
2.7.3

Re: i915 drm patches in LTS 4.4.x ?

[gregkh]

On Mon, Jul 25, 2016 at 12:01:00PM +0000, Joakim Tjernlund wrote:
> On Mon, 2016-07-25 at 13:39 +0200, Joakim Tjernlund wrote:
> > On Sun, 2016-07-24 at 16:41 -0700, gregkh@linuxfoundation.org wrote:
> > > 
> > > On Wed, Jul 20, 2016 at 08:41:06AM +0000, Joakim Tjernlund wrote:
> > > > 
> > > > 
> > > > On Tue, 2016-07-19 at 15:06 -0700, gregkh@linuxfoundation.org wrote:
> > ....
> > 
> > > 
> > > > 
> > > > > 
> > > > > �which has a problem with external monitors,
> > > > > > 
> > > > > > they go blank/off intermittently. I think/hope this has been fixed upstream by now but I not sure.
> > > > > > One path caught my eye though was this:
> > > > > > � drm/i915: Revert DisplayPort fast link training feature
> > > > > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/i915?id=34511
> > > > > > dc
> > > > > > e4b3
> > > > > > 5685d3988d5c8b100d11a068db5bd
> > > > > > 
> > > > > > Is that one in your queue?
> > > > > 
> > > > > Yes it is, I have not gotten to that one yet.
> > > > > �
> > > > 
> > > > I did a quick backport of above patch for 4.4 and had a guy test it his laptop and it did
> > > > fix the problem :)
> > > 
> > > Can you provide the backport for 4.4 and 4.6-stable trees, as it
> > > currently does not apply for me at all :(
> > > �
> > 
> > Just did a backport to 4.4( I hope it applies to 4.6 as well:) but git send email does not work
> > ATM because we are moving to a new mail domain and it is not finished yet. Attaching the patch instead.
> 
> Well, it did not apply to 4.6.y so here is backport

Thanks for both of these patches, now queued up.

greg k-h