transitioning from 4.17.18 to 5.0.6

transitioning from 4.17.18 to 5.0.6

[Patrick Venture]

HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
with the quanta-q71l board.  More specifically now my host can't talk
to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
can talk to ipmid file via dbus, and btbridge is running.

Has anyone else run into weirdness?

I am seeing:

[    4.645814] ipmi-bt-host 1e789140.ibt: Found bt bmc device
[    4.650341] ipmi-bt-host 1e789140.ibt: Using IRQ 32

The dts file I'm using is basically identical to the upstream one, and
my defconfig is also nearly identical to what's in meta-aspeed for the
ast2400.

Patrick

Re: transitioning from 4.17.18 to 5.0.6

[Patrick Venture]

On Tue, Apr 9, 2019 at 9:35 AM Patrick Venture <venture@google.com> wrote:
>
> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> with the quanta-q71l board.  More specifically now my host can't talk
> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> can talk to ipmid file via dbus, and btbridge is running.
>
> Has anyone else run into weirdness?
>
> I am seeing:
>
> [    4.645814] ipmi-bt-host 1e789140.ibt: Found bt bmc device
> [    4.650341] ipmi-bt-host 1e789140.ibt: Using IRQ 32
>
> The dts file I'm using is basically identical to the upstream one, and
> my defconfig is also nearly identical to what's in meta-aspeed for the
> ast2400.

Verified that with the same software stack (reverted the
i2c-path-hwmon-config-change) works fine on 4.17.18.

>
> Patrick

Re: transitioning from 4.17.18 to 5.0.6

[Joel Stanley]

On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
>
> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> with the quanta-q71l board.  More specifically now my host can't talk
> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> can talk to ipmid file via dbus, and btbridge is running.
>
> Has anyone else run into weirdness?

I've not seen that.

My only guess is there is an issue with clocking. Can you try booting
the machine with clk_ignore_unused in your kernel command line?

>
> I am seeing:
>
> [    4.645814] ipmi-bt-host 1e789140.ibt: Found bt bmc device
> [    4.650341] ipmi-bt-host 1e789140.ibt: Using IRQ 32
>
> The dts file I'm using is basically identical to the upstream one, and
> my defconfig is also nearly identical to what's in meta-aspeed for the
> ast2400.
>
> Patrick

Re: transitioning from 4.17.18 to 5.0.6

[Patrick Venture]

On Tue, Apr 9, 2019 at 7:17 PM Joel Stanley <joel@jms.id.au> wrote:
>
> On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
> >
> > HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> > with the quanta-q71l board.  More specifically now my host can't talk
> > to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> > can talk to ipmid file via dbus, and btbridge is running.
> >
> > Has anyone else run into weirdness?
>
> I've not seen that.
>
> My only guess is there is an issue with clocking. Can you try booting
> the machine with clk_ignore_unused in your kernel command line?

I'll give it a shot, although IPMI not working is odd since the
LPC_CLK is enabled.  But yeah, I'll try that today and reply back,
thanks.

>
> >
> > I am seeing:
> >
> > [    4.645814] ipmi-bt-host 1e789140.ibt: Found bt bmc device
> > [    4.650341] ipmi-bt-host 1e789140.ibt: Using IRQ 32
> >
> > The dts file I'm using is basically identical to the upstream one, and
> > my defconfig is also nearly identical to what's in meta-aspeed for the
> > ast2400.
> >
> > Patrick

Re: transitioning from 4.17.18 to 5.0.6

[Cédric Le Goater]

On 4/10/19 4:39 PM, Patrick Venture wrote:
> On Tue, Apr 9, 2019 at 7:17 PM Joel Stanley <joel@jms.id.au> wrote:
>>
>> On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
>>>
>>> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
>>> with the quanta-q71l board.  More specifically now my host can't talk
>>> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
>>> can talk to ipmid file via dbus, and btbridge is running.
>>>
>>> Has anyone else run into weirdness?
>>
>> I've not seen that.
>>
>> My only guess is there is an issue with clocking. Can you try booting
>> the machine with clk_ignore_unused in your kernel command line?
> 
> I'll give it a shot, although IPMI not working is odd since the
> LPC_CLK is enabled.  But yeah, I'll try that today and reply back,
> thanks.

I don't see this issue on a palmetto, BMC running with 5.0.5.

C. 


> 
>>
>>>
>>> I am seeing:
>>>
>>> [    4.645814] ipmi-bt-host 1e789140.ibt: Found bt bmc device
>>> [    4.650341] ipmi-bt-host 1e789140.ibt: Using IRQ 32
>>>
>>> The dts file I'm using is basically identical to the upstream one, and
>>> my defconfig is also nearly identical to what's in meta-aspeed for the
>>> ast2400.
>>>
>>> Patrick

Re: transitioning from 4.17.18 to 5.0.6

[Patrick Venture]

On Wed, Apr 10, 2019 at 9:02 AM Cédric Le Goater <clg@kaod.org> wrote:
>
> On 4/10/19 4:39 PM, Patrick Venture wrote:
> > On Tue, Apr 9, 2019 at 7:17 PM Joel Stanley <joel@jms.id.au> wrote:
> >>
> >> On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
> >>>
> >>> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> >>> with the quanta-q71l board.  More specifically now my host can't talk
> >>> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> >>> can talk to ipmid file via dbus, and btbridge is running.
> >>>
> >>> Has anyone else run into weirdness?
> >>
> >> I've not seen that.
> >>
> >> My only guess is there is an issue with clocking. Can you try booting
> >> the machine with clk_ignore_unused in your kernel command line?
> >
> > I'll give it a shot, although IPMI not working is odd since the
> > LPC_CLK is enabled.  But yeah, I'll try that today and reply back,
> > thanks.

root@localhost:~# cat /proc/cmdline
console=ttyS4,115200n8 root=/dev/ram rw clk_ignore_unused

./gbmctool -command firmware -action get -field version
Get Device ID command failed: 0xd5 Command not supported in present state
E0410 10:12:01.640883  456408 real_main.cc:53] Unable to send code:
Command not supported in present state

So that didn't fix it unless I need to reboot the host as well.

>
> I don't see this issue on a palmetto, BMC running with 5.0.5.
>
> C.
>
>
> >
> >>
> >>>
> >>> I am seeing:
> >>>
> >>> [    4.645814] ipmi-bt-host 1e789140.ibt: Found bt bmc device
> >>> [    4.650341] ipmi-bt-host 1e789140.ibt: Using IRQ 32
> >>>
> >>> The dts file I'm using is basically identical to the upstream one, and
> >>> my defconfig is also nearly identical to what's in meta-aspeed for the
> >>> ast2400.
> >>>
> >>> Patrick
>

Re: transitioning from 4.17.18 to 5.0.6

[Patrick Venture]

On Wed, Apr 10, 2019 at 10:13 AM Patrick Venture <venture@google.com> wrote:
>
> On Wed, Apr 10, 2019 at 9:02 AM Cédric Le Goater <clg@kaod.org> wrote:
> >
> > On 4/10/19 4:39 PM, Patrick Venture wrote:
> > > On Tue, Apr 9, 2019 at 7:17 PM Joel Stanley <joel@jms.id.au> wrote:
> > >>
> > >> On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
> > >>>
> > >>> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> > >>> with the quanta-q71l board.  More specifically now my host can't talk
> > >>> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> > >>> can talk to ipmid file via dbus, and btbridge is running.
> > >>>
> > >>> Has anyone else run into weirdness?
> > >>
> > >> I've not seen that.
> > >>
> > >> My only guess is there is an issue with clocking. Can you try booting
> > >> the machine with clk_ignore_unused in your kernel command line?
> > >
> > > I'll give it a shot, although IPMI not working is odd since the
> > > LPC_CLK is enabled.  But yeah, I'll try that today and reply back,
> > > thanks.
>
> root@localhost:~# cat /proc/cmdline
> console=ttyS4,115200n8 root=/dev/ram rw clk_ignore_unused
>
> ./gbmctool -command firmware -action get -field version
> Get Device ID command failed: 0xd5 Command not supported in present state
> E0410 10:12:01.640883  456408 real_main.cc:53] Unable to send code:
> Command not supported in present state
>
> So that didn't fix it unless I need to reboot the host as well.
>
> >
> > I don't see this issue on a palmetto, BMC running with 5.0.5.

Looks like the issue is 5.0.6 doesn't have the lpc parameter optional patchset.

> >
> > C.
> >
> >
> > >
> > >>
> > >>>
> > >>> I am seeing:
> > >>>
> > >>> [    4.645814] ipmi-bt-host 1e789140.ibt: Found bt bmc device
> > >>> [    4.650341] ipmi-bt-host 1e789140.ibt: Using IRQ 32
> > >>>
> > >>> The dts file I'm using is basically identical to the upstream one, and
> > >>> my defconfig is also nearly identical to what's in meta-aspeed for the
> > >>> ast2400.
> > >>>
> > >>> Patrick
> >

Re: transitioning from 4.17.18 to 5.0.6

[Andrew Jeffery]

On Tue, 16 Apr 2019, at 01:06, Patrick Venture wrote:
> On Wed, Apr 10, 2019 at 10:13 AM Patrick Venture <venture@google.com> wrote:
> >
> > On Wed, Apr 10, 2019 at 9:02 AM Cédric Le Goater <clg@kaod.org> wrote:
> > >
> > > On 4/10/19 4:39 PM, Patrick Venture wrote:
> > > > On Tue, Apr 9, 2019 at 7:17 PM Joel Stanley <joel@jms.id.au> wrote:
> > > >>
> > > >> On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
> > > >>>
> > > >>> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> > > >>> with the quanta-q71l board.  More specifically now my host can't talk
> > > >>> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> > > >>> can talk to ipmid file via dbus, and btbridge is running.
> > > >>>
> > > >>> Has anyone else run into weirdness?
> > > >>
> > > >> I've not seen that.
> > > >>
> > > >> My only guess is there is an issue with clocking. Can you try booting
> > > >> the machine with clk_ignore_unused in your kernel command line?
> > > >
> > > > I'll give it a shot, although IPMI not working is odd since the
> > > > LPC_CLK is enabled.  But yeah, I'll try that today and reply back,
> > > > thanks.
> >
> > root@localhost:~# cat /proc/cmdline
> > console=ttyS4,115200n8 root=/dev/ram rw clk_ignore_unused
> >
> > ./gbmctool -command firmware -action get -field version
> > Get Device ID command failed: 0xd5 Command not supported in present state
> > E0410 10:12:01.640883  456408 real_main.cc:53] Unable to send code:
> > Command not supported in present state
> >
> > So that didn't fix it unless I need to reboot the host as well.
> >
> > >
> > > I don't see this issue on a palmetto, BMC running with 5.0.5.
> 
> Looks like the issue is 5.0.6 doesn't have the lpc parameter optional patchset.

The only place I've found the patch applied is in one of Joel's personal trees.

Seems bizarre.

Andrew

Re: transitioning from 4.17.18 to 5.0.6

[Patrick Venture]

On Mon, Apr 15, 2019 at 8:05 PM Andrew Jeffery <andrew@aj.id.au> wrote:
>
> On Tue, 16 Apr 2019, at 01:06, Patrick Venture wrote:
> > On Wed, Apr 10, 2019 at 10:13 AM Patrick Venture <venture@google.com> wrote:
> > >
> > > On Wed, Apr 10, 2019 at 9:02 AM Cédric Le Goater <clg@kaod.org> wrote:
> > > >
> > > > On 4/10/19 4:39 PM, Patrick Venture wrote:
> > > > > On Tue, Apr 9, 2019 at 7:17 PM Joel Stanley <joel@jms.id.au> wrote:
> > > > >>
> > > > >> On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
> > > > >>>
> > > > >>> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> > > > >>> with the quanta-q71l board.  More specifically now my host can't talk
> > > > >>> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> > > > >>> can talk to ipmid file via dbus, and btbridge is running.
> > > > >>>
> > > > >>> Has anyone else run into weirdness?
> > > > >>
> > > > >> I've not seen that.
> > > > >>
> > > > >> My only guess is there is an issue with clocking. Can you try booting
> > > > >> the machine with clk_ignore_unused in your kernel command line?
> > > > >
> > > > > I'll give it a shot, although IPMI not working is odd since the
> > > > > LPC_CLK is enabled.  But yeah, I'll try that today and reply back,
> > > > > thanks.
> > >
> > > root@localhost:~# cat /proc/cmdline
> > > console=ttyS4,115200n8 root=/dev/ram rw clk_ignore_unused
> > >
> > > ./gbmctool -command firmware -action get -field version
> > > Get Device ID command failed: 0xd5 Command not supported in present state
> > > E0410 10:12:01.640883  456408 real_main.cc:53] Unable to send code:
> > > Command not supported in present state
> > >
> > > So that didn't fix it unless I need to reboot the host as well.
> > >
> > > >
> > > > I don't see this issue on a palmetto, BMC running with 5.0.5.
> >
> > Looks like the issue is 5.0.6 doesn't have the lpc parameter optional patchset.
>
> The only place I've found the patch applied is in one of Joel's personal trees.

That would explain why I couldn't find it in the usual places.

>
> Seems bizarre.

Sorry, I stopped following that patchset a while back, I just assumed
it landed.  Was there some push back?

>
> Andrew

Re: transitioning from 4.17.18 to 5.0.6

[Andrew Jeffery]

On Tue, 16 Apr 2019, at 23:56, Patrick Venture wrote:
> On Mon, Apr 15, 2019 at 8:05 PM Andrew Jeffery <andrew@aj.id.au> wrote:
> >
> > On Tue, 16 Apr 2019, at 01:06, Patrick Venture wrote:
> > > On Wed, Apr 10, 2019 at 10:13 AM Patrick Venture <venture@google.com> wrote:
> > > >
> > > > On Wed, Apr 10, 2019 at 9:02 AM Cédric Le Goater <clg@kaod.org> wrote:
> > > > >
> > > > > On 4/10/19 4:39 PM, Patrick Venture wrote:
> > > > > > On Tue, Apr 9, 2019 at 7:17 PM Joel Stanley <joel@jms.id.au> wrote:
> > > > > >>
> > > > > >> On Tue, 9 Apr 2019 at 16:35, Patrick Venture <venture@google.com> wrote:
> > > > > >>>
> > > > > >>> HI, I ran into some weirdness when transitioning from 4.17.18 to 5.0.6
> > > > > >>> with the quanta-q71l board.  More specifically now my host can't talk
> > > > > >>> to the BMC via ipmi-bt.  All it returns is 0xd5.  on the BMC itself, I
> > > > > >>> can talk to ipmid file via dbus, and btbridge is running.
> > > > > >>>
> > > > > >>> Has anyone else run into weirdness?
> > > > > >>
> > > > > >> I've not seen that.
> > > > > >>
> > > > > >> My only guess is there is an issue with clocking. Can you try booting
> > > > > >> the machine with clk_ignore_unused in your kernel command line?
> > > > > >
> > > > > > I'll give it a shot, although IPMI not working is odd since the
> > > > > > LPC_CLK is enabled.  But yeah, I'll try that today and reply back,
> > > > > > thanks.
> > > >
> > > > root@localhost:~# cat /proc/cmdline
> > > > console=ttyS4,115200n8 root=/dev/ram rw clk_ignore_unused
> > > >
> > > > ./gbmctool -command firmware -action get -field version
> > > > Get Device ID command failed: 0xd5 Command not supported in present state
> > > > E0410 10:12:01.640883  456408 real_main.cc:53] Unable to send code:
> > > > Command not supported in present state
> > > >
> > > > So that didn't fix it unless I need to reboot the host as well.
> > > >
> > > > >
> > > > > I don't see this issue on a palmetto, BMC running with 5.0.5.
> > >
> > > Looks like the issue is 5.0.6 doesn't have the lpc parameter optional patchset.
> >
> > The only place I've found the patch applied is in one of Joel's personal trees.
> 
> That would explain why I couldn't find it in the usual places.
> 
> >
> > Seems bizarre.
> 
> Sorry, I stopped following that patchset a while back, I just assumed
> it landed.  Was there some push back?

Not that I'm aware of. Vijay has been pinging Joel periodically, I'm not
sure why he hasn't merged it. I'd like to wait until Joel can respond in
case there's something I've missed here.

Andrew

Re: transitioning from 4.17.18 to 5.0.6

[Joel Stanley]

On Thu, 18 Apr 2019 at 06:20, Andrew Jeffery <andrew@aj.id.au> wrote:
> On Tue, 16 Apr 2019, at 23:56, Patrick Venture wrote:
> > > > Looks like the issue is 5.0.6 doesn't have the lpc parameter optional patchset.
> > Sorry, I stopped following that patchset a while back, I just assumed
> > it landed.  Was there some push back?
>
> Not that I'm aware of. Vijay has been pinging Joel periodically, I'm not
> sure why he hasn't merged it. I'd like to wait until Joel can respond in
> case there's something I've missed here.

I can't see anything relevant on the list. In fact, I've cleared the
entire backlog today.

$ pwclient list -p openbmc -s new  | grep dev-5.0 | wc -l
0

Do you have a link to the email in question? If not, someone please
send it along and I'll put it in the tree. I assumed it was applied.

Cheers,

Joel

Re: transitioning from 4.17.18 to 5.0.6

[Andrew Jeffery]

On Mon, 29 Apr 2019, at 17:57, Joel Stanley wrote:
> On Thu, 18 Apr 2019 at 06:20, Andrew Jeffery <andrew@aj.id.au> wrote:
> > On Tue, 16 Apr 2019, at 23:56, Patrick Venture wrote:
> > > > > Looks like the issue is 5.0.6 doesn't have the lpc parameter optional patchset.
> > > Sorry, I stopped following that patchset a while back, I just assumed
> > > it landed.  Was there some push back?
> >
> > Not that I'm aware of. Vijay has been pinging Joel periodically, I'm not
> > sure why he hasn't merged it. I'd like to wait until Joel can respond in
> > case there's something I've missed here.
> 
> I can't see anything relevant on the list. In fact, I've cleared the
> entire backlog today.
> 
> $ pwclient list -p openbmc -s new  | grep dev-5.0 | wc -l
> 0
> 
> Do you have a link to the email in question? If not, someone please
> send it along and I'll put it in the tree. I assumed it was applied.

Maybe it's best if Vijay resends it. Add him in To.

Andrew

> 
> Cheers,
> 
> Joel
>