Port-arm archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: aarch64{,eb} fails to boot on RPI3 (Re: 2021-06-18-netbsd-raspi-earmv6hf.img (Re: Raspberry Pi update please.))



Increasing PERCPU_IMPORT_SIZE may solve this issue.

2021年6月21日(月) 19:08 Rin Okuyama <rokuyama.rk%gmail.com@localhost>:
>
> Hi,
>
> On 2021/06/21 8:07, Paul Goyette wrote:
> > On Fri, 18 Jun 2021, Paul Goyette wrote:
> >
> >> This might be mine - I will work with Rin to investigate.  I don't
> >> have any arm hardware, so will likely need help with testing any
> >> possible fix.
> >
> > I'm looked pretty hard at this, and I can only guess that my commit
> > simply exposed an existing bug in the arm code.  Since I know nearly
> > nothing about arm, I don't have any clues on where to go from here.
> >
> > Hopefully someone with arm knowledge can dig deeper into the root
> > cause.
>
> Paul, thank you very much for investigation. And I'm sorry that I was
> unavailable in the weekend; I had a network trouble in my home.
>
> I've found that 32-bit kernels are also affected. Now, I think that
> this is a problem for bcm2836mp_intr_init().
>
> I send PR port-arm/56264 for this:
>
> http://gnats.netbsd.org/56264
>
> Thanks,
> rin
>
> >> On Fri, 18 Jun 2021, Rin Okuyama wrote:
> >>
> >>> On 2021/06/18 18:05, Rin Okuyama wrote:
> >>>> Hi,
> >>>>
> >>>> On 2021/06/18 8:59, john%ziaspace.com@localhost wrote:
> >>>>>> Update:
> >>>>>> - NetBSD 9.99.85 evbarm-earmv6hf 202106170850Z rpi.img from nyftp.
> >>>>>>
> >>>>>> - Raspberry Pi [0-3] have been supported in big-endian mode (Rin Okuyama)
> >>>>>> XXX: should test big-endian image and pkgsrc
> >>>>>
> >>>>> earmv7hfeb works fine so far, pkgsrc included. aarc64eb doesn't boot on a Raspberry Pi 3, but I'll send more details about that when I collect them.
> >>>>
> >>>> Hmm, both aarch64{,eb} fails to boot on RPI3:
> >>>>
> >>>> ----
> >>>> [   1.3505790] panic: kernel diagnostic assertion "l->l_stat == LSONPROC" failed: file "../../../../kern/kern_sleepq.c", line 227
> >>>> [   1.3705814] cpu1: Begin traceback...
> >>>> [   1.3705814] trace fp ffffc0006eba79b0
> >>>> [   1.3805840] fp ffffc0006eba79e0 vpanic() at ffffc000005561ec netbsd:vpanic+0x14c
> >>>> [   1.3905924] fp ffffc0006eba7a40 kern_assert() at ffffc000007b3e68 netbsd:kern_assert+0x58
> >>>> [   1.4105922] fp ffffc0006eba7ad0 sleepq_enqueue() at ffffc0000051ea54 netbsd:sleepq_enqueue+0x174
> >>>> [   1.4206018] fp ffffc0006eba7b10 cv_enter() at ffffc000004e5080 netbsd:cv_enter+0xf0
> >>>> [   1.4406016] fp ffffc0006eba7b50 cv_wait() at ffffc000004e52a8 netbsd:cv_wait+0x38
> >>>> [   1.4506010] fp ffffc0006eba7b80 xc_wait() at ffffc0000055fbb0 netbsd:xc_wait+0xb0
> >>>> [   1.4706047] fp ffffc0006eba7bc0 percpu_backend_alloc() at ffffc0000054cf14 netbsd:percpu_backend_alloc+0x154
> >>>> [   1.4906100] fp ffffc0006eba7c40 vmem_xalloc() at ffffc0000055dcd8 netbsd:vmem_xalloc+0x578
> >>>> [   1.5106159] fp ffffc0006eba7d10 vmem_alloc() at ffffc0000055e1e4 netbsd:vmem_alloc+0x84
> >>>> [   1.5206190] fp ffffc0006eba7d70 percpu_create() at ffffc0000054d3e0 netbsd:percpu_create+0x40
> >>>> [   1.5406215] fp ffffc0006eba7e00 pic_add() at ffffc0000000314c netbsd:pic_add+0xfc
> >>>> [   1.5606308] fp ffffc0006eba7e40 bcm2836mp_intr_init() at ffffc0000001adb0 netbsd:bcm2836mp_intr_init+0x90
> >>>> [   1.5706315] fp ffffc0006eba7e90 arm_fdt_cpu_hatch() at ffffc0000006a2f8 netbsd:arm_fdt_cpu_hatch+0x24
> >>>> [   1.5906332] fp ffffc0006eba7eb0 cpu_hatch() at ffffc0000009c29c netbsd:cpu_hatch+0xbc
> >>>> [   1.6106367] fp 0000000000000000 cpu_mpstart() at ffffc00000001a88 netbsd:cpu_mpstart+0x19c
> >>>> [   1.6306431] cpu1: End traceback...
> >>>> ----
> >>>>
> >>>> For my RPI3, kernel as of 2020-05-14 is working:
> >>>>
> >>>> NetBSD 9.99.82 (GENERIC64) #0: Fri May 14 16:32:10 JST 2021
> >>>>
> >>>> I will bisect the problem...
> >>>
> >>> I found that the failure starts with this commit:
> >>>
> >>> http://www.nerv.org/netbsd/changeset.cgi?id=20210609T232251Z.095141ac330a067f4506a48b81678d3ed2bf99f9#src/sys/dev/dev_verbose.h
> >>>
> >>> If dev_verbose.h is reverted to rev 1.5, both aarch64{,eb} successfully
> >>> boot on RPI3. But, I still don't understand why this change causes the
> >>> panic.
> >>>
> >>> Also, I found PR which may be related to:
> >>>
> >>> http://gnats.netbsd.org/55889
> >>>
> >>> I will examine further tomorrow...
> >>>
> >>> Thanks,
> >>> rin
> >>>
> >>>
> >>>
> >>>
> >>
> >> +--------------------+--------------------------+----------------------+
> >> | Paul Goyette       | PGP Key fingerprint:     | E-mail addresses:    |
> >> | (Retired)          | FA29 0E3B 35AF E8AE 6651 | paul%whooppee.com@localhost    |
> >> | Software Developer | 0786 F758 55DE 53BA 7731 | pgoyette%netbsd.org@localhost  |
> >> |                    |                          | pgoyette99%gmail.com@localhost |
> >> +--------------------+--------------------------+----------------------+
> >>
> >> !DSPAM:60ccb31225151735754203!
> >>
> >
> > +--------------------+--------------------------+----------------------+
> > | Paul Goyette       | PGP Key fingerprint:     | E-mail addresses:    |
> > | (Retired)          | FA29 0E3B 35AF E8AE 6651 | paul%whooppee.com@localhost    |
> > | Software Developer | 0786 F758 55DE 53BA 7731 | pgoyette%netbsd.org@localhost  |
> > |                    |                          | pgoyette99%gmail.com@localhost |
> > +--------------------+--------------------------+----------------------+


Home | Main Index | Thread Index | Old Index