[Letux-kernel] LetuxOS: Kernel: m-latest 5.7-rc6 successful

Stefan Leichter sle85276 at gmx.de
Mon Jun 1 23:47:25 CEST 2020


Hello Nikolaus,

Am 25.05.20 um 16:59 schrieb H. Nikolaus Schaller:
> Hi Stefan,
>
>> Am 24.05.2020 um 22:25 schrieb Stefan Leichter <sle85276 at gmx.de>:
>>
>> Am 24.05.20 um 14:51 schrieb H. Nikolaus Schaller:
>>> Hi Stefan,
>>>
>>>
>>
>> Hello Nikolaus,
>>
>> with the reverted commit 01399e6be1eab63dbd1b9f1f653f74fc8beec8f2 the Pandora boots to the login prompt too.
>
> Ok, good.
>
> Of course it has probably no display then, because there are two display channels defined.
> If you want to play or verify you could remove
>
>                 display1 = &tv;
>
> from arch/arm/boot/dts/omap3-pandora-common.dtsi
>
>>
>> Attached are the console logs from the 5.7-rc3 kernel from your sever and from my self build kernel with the reverted commit.
>>
>> Best regards
>> Stefan
>> <console-5.7.0-rc3.log.bz2><console-5.7.0-rc3-revert.log.bz2>
>
> Fine!
>
> I have looked through it and there is no obvious error message
> related to the tvout or omapdss.
>
> Instead it is well initialized
>
> [   10.753173] [drm] Initialized omapdrm 1.0.0 20110917 for omapdrm.0 on minor 1
> [   11.766357] drm_setup_crtcs_fb: sw_rotations=1
> [   11.772399] drm_setup_crtcs_fb: fbcon_rotate_hint=0
>
> and the problem starts later. So it *could* be that the tvout is just
> interfering and triggering some other bug.
>
> What I also observe is that the drivers/base/dd.c warning comes much later
> with your reverted build! This indicates a race condition between multiple
> activities during boot.
>
> It also may be that some driver is accessing out-of bounds array.
> I already had that some years ago and it took weeks to track it down.
>
> For me the drivers/base/dd.c problem has unexpectedly disappeared with
> letux-5.7-rc6 with no known reason or change.
>
> Maybe you can give letux-5.7-rc6 a try? Either downloaded binary or
> self-compiled w/o or with tvout revert.
>
> Best regards,
> Nikolaus
> Finally I found that the revert of 01399e6be1eab63dbd1b9f1f653f74fc8beec8f2 does
not solve the boot hang reliable. Therefore I re-did the git bisect on 5.7-rc7 and
booted the kernel up to 10 times before deciding about good or bad.

This brings me to the commit 5bd99ae74ef584bc56895ead2ad070381767ff42 as first
bad commit.

Reverting the commit make the kernel 5.7-rc7 boot reliable for me.

Best regards
Stefan


More information about the Letux-kernel mailing list