From: Zhengyuan Liu <hidden> Date: 2021-01-11 08:27:07
Hi, all
When boot v5.11-rc2 on my Kunpeng 920 PC, the kernel will block at PMU
initialization stage, and after binary-search with bisect I found that
this problem was introduced by commit ac4511c9364c ("drivers/perf:
hisi: Add identifier sysfs file"),
and to be more accurate it actually blocked at following position:
hisi_hha_pmu_init_data():
hha_pmu->identifier = readl(hha_pmu->base + HHA_VERSION);
If you guys had already know this problem, please ignore this message.
Thanks,
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
From: John Garry <hidden> Date: 2021-01-11 09:17:50
On 11/01/2021 08:25, Zhengyuan Liu wrote:
Hi, all
When boot v5.11-rc2 on my Kunpeng 920 PC, the kernel will block at PMU
initialization stage, and after binary-search with bisect I found that
this problem was introduced by commit ac4511c9364c ("drivers/perf:
hisi: Add identifier sysfs file"),
and to be more accurate it actually blocked at following position:
hisi_hha_pmu_init_data():
hha_pmu->identifier = readl(hha_pmu->base + HHA_VERSION);
If you guys had already know this problem, please ignore this message.
Do you have a dmesg? I don't see anything to block here, only crash.
Does v5.11-rc1 boot ok?
It works fine on my D06CS (which includes Kunpeng 920) with rc3:
john@debian:~$ more
/sys/bus/event_source/devices/hisi_sccl5_hha7/identifier
0x00000000
john@debian:~$ uname -a
Linux debian 5.11.0-rc3-dirty #139 SMP PREEMPT Mon Jan 11 08:42:26 GMT
2021 aarch64 GNU/Linux
john@debian:~$
@Shaokun, Do you know what could be the problem on PC?
Thanks,
John
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Hi John/Zhengyuan,
在 2021/1/11 17:13, John Garry 写道:
On 11/01/2021 08:25, Zhengyuan Liu wrote:
quoted
Hi, all
When boot v5.11-rc2 on my Kunpeng 920 PC, the kernel will block at PMU
initialization stage, and after binary-search with bisect I found that
this problem was introduced by commit ac4511c9364c ("drivers/perf:
hisi: Add identifier sysfs file"),
and to be more accurate it actually blocked at following position:
hisi_hha_pmu_init_data():
hha_pmu->identifier = readl(hha_pmu->base + HHA_VERSION);
If you guys had already know this problem, please ignore this message.
Do you have a dmesg? I don't see anything to block here, only crash.
Does v5.11-rc1 boot ok?
It works fine on my D06CS (which includes Kunpeng 920) with rc3:
john@debian:~$ more /sys/bus/event_source/devices/hisi_sccl5_hha7/identifier
0x00000000
john@debian:~$ uname -a
Linux debian 5.11.0-rc3-dirty #139 SMP PREEMPT Mon Jan 11 08:42:26 GMT 2021 aarch64 GNU/Linux
john@debian:~$
@Shaokun, Do you know what could be the problem on PC?
I doubt the UEFI which maybe give more HHA numbers than PC has or give the wrong
mmap address space that may cause this issue.
So please can you tell more information about the UEFI?
Shaokun,
From: Zhengyuan Liu <hidden> Date: 2021-01-11 11:43:08
On Mon, Jan 11, 2021 at 5:34 PM Shaokun Zhang
[off-list ref] wrote:
Hi John/Zhengyuan,
在 2021/1/11 17:13, John Garry 写道:
quoted
On 11/01/2021 08:25, Zhengyuan Liu wrote:
quoted
Hi, all
When boot v5.11-rc2 on my Kunpeng 920 PC, the kernel will block at PMU
initialization stage, and after binary-search with bisect I found that
this problem was introduced by commit ac4511c9364c ("drivers/perf:
hisi: Add identifier sysfs file"),
and to be more accurate it actually blocked at following position:
hisi_hha_pmu_init_data():
hha_pmu->identifier = readl(hha_pmu->base + HHA_VERSION);
If you guys had already know this problem, please ignore this message.
Do you have a dmesg? I don't see anything to block here, only crash.
Does v5.11-rc1 boot ok?
V5.11-rc1 has included this patch, therefore also bad.
The dmesg does not have any error message, It only blocked which seems
the cpu doesn't work,
and I can hear the rotation rate of fans increasing rapidly.
I also built the PMU from Y to M, it has the same phenomenon when the
module is loaded.
quoted
It works fine on my D06CS (which includes Kunpeng 920) with rc3:
john@debian:~$ more /sys/bus/event_source/devices/hisi_sccl5_hha7/identifier
0x00000000
john@debian:~$ uname -a
Linux debian 5.11.0-rc3-dirty #139 SMP PREEMPT Mon Jan 11 08:42:26 GMT 2021 aarch64 GNU/Linux
john@debian:~$
@Shaokun, Do you know what could be the problem on PC?
I doubt the UEFI which maybe give more HHA numbers than PC has or give the wrong
mmap address space that may cause this issue.
So please can you tell more information about the UEFI?
Hi Zhengyuan,
Thanks for your quick reply,
在 2021/1/11 19:38, Zhengyuan Liu 写道:
On Mon, Jan 11, 2021 at 5:34 PM Shaokun Zhang
[off-list ref] wrote:
quoted
Hi John/Zhengyuan,
在 2021/1/11 17:13, John Garry 写道:
quoted
On 11/01/2021 08:25, Zhengyuan Liu wrote:
quoted
Hi, all
When boot v5.11-rc2 on my Kunpeng 920 PC, the kernel will block at PMU
initialization stage, and after binary-search with bisect I found that
this problem was introduced by commit ac4511c9364c ("drivers/perf:
hisi: Add identifier sysfs file"),
and to be more accurate it actually blocked at following position:
hisi_hha_pmu_init_data():
hha_pmu->identifier = readl(hha_pmu->base + HHA_VERSION);
If you guys had already know this problem, please ignore this message.
Do you have a dmesg? I don't see anything to block here, only crash.
Does v5.11-rc1 boot ok?
V5.11-rc1 has included this patch, therefore also bad.
The dmesg does not have any error message, It only blocked which seems
the cpu doesn't work,
and I can hear the rotation rate of fans increasing rapidly.
I also built the PMU from Y to M, it has the same phenomenon when the
module is loaded.
quoted
quoted
It works fine on my D06CS (which includes Kunpeng 920) with rc3:
john@debian:~$ more /sys/bus/event_source/devices/hisi_sccl5_hha7/identifier
0x00000000
john@debian:~$ uname -a
Linux debian 5.11.0-rc3-dirty #139 SMP PREEMPT Mon Jan 11 08:42:26 GMT 2021 aarch64 GNU/Linux
john@debian:~$
@Shaokun, Do you know what could be the problem on PC?
I doubt the UEFI which maybe give more HHA numbers than PC has or give the wrong
mmap address space that may cause this issue.
So please can you tell more information about the UEFI?
Did you mean DSDT or Bios Information?
I have checked that there are some wrong PMU nodes enumerating in the UEFI
file and are you happy to get the new UEFI file from Xinwei and test it again?
Apologies for this mistake.
Thanks,
Shaokun
From: Zhengyuan Liu <hidden> Date: 2021-01-12 07:48:24
On Tue, Jan 12, 2021 at 3:39 PM Shaokun Zhang
[off-list ref] wrote:
Hi Zhengyuan,
Thanks for your quick reply,
在 2021/1/11 19:38, Zhengyuan Liu 写道:
quoted
On Mon, Jan 11, 2021 at 5:34 PM Shaokun Zhang
[off-list ref] wrote:
quoted
Hi John/Zhengyuan,
在 2021/1/11 17:13, John Garry 写道:
quoted
On 11/01/2021 08:25, Zhengyuan Liu wrote:
quoted
Hi, all
When boot v5.11-rc2 on my Kunpeng 920 PC, the kernel will block at PMU
initialization stage, and after binary-search with bisect I found that
this problem was introduced by commit ac4511c9364c ("drivers/perf:
hisi: Add identifier sysfs file"),
and to be more accurate it actually blocked at following position:
hisi_hha_pmu_init_data():
hha_pmu->identifier = readl(hha_pmu->base + HHA_VERSION);
If you guys had already know this problem, please ignore this message.
Do you have a dmesg? I don't see anything to block here, only crash.
Does v5.11-rc1 boot ok?
V5.11-rc1 has included this patch, therefore also bad.
The dmesg does not have any error message, It only blocked which seems
the cpu doesn't work,
and I can hear the rotation rate of fans increasing rapidly.
I also built the PMU from Y to M, it has the same phenomenon when the
module is loaded.
quoted
quoted
It works fine on my D06CS (which includes Kunpeng 920) with rc3:
john@debian:~$ more /sys/bus/event_source/devices/hisi_sccl5_hha7/identifier
0x00000000
john@debian:~$ uname -a
Linux debian 5.11.0-rc3-dirty #139 SMP PREEMPT Mon Jan 11 08:42:26 GMT 2021 aarch64 GNU/Linux
john@debian:~$
@Shaokun, Do you know what could be the problem on PC?
I doubt the UEFI which maybe give more HHA numbers than PC has or give the wrong
mmap address space that may cause this issue.
So please can you tell more information about the UEFI?
Did you mean DSDT or Bios Information?
I have checked that there are some wrong PMU nodes enumerating in the UEFI
file and are you happy to get the new UEFI file from Xinwei and test it again?
Yes, of course, I'll use the new UEFI to test again and do a feedback later.
Thanks,