On 4/22/22 12:16, Arnd Bergmann wrote:
On Fri, Apr 22, 2022 at 7:05 PM Guenter Roeck linux@roeck-us.net wrote:
On Tue, Apr 19, 2022 at 06:37:22PM +0200, Arnd Bergmann wrote:
From: Arnd Bergmann arnd@arndb.de
This revisits a series I sent a few years ago:
https://lore.kernel.org/lkml/20191018154052.1276506-1-arnd@arndb.de/
All the other ARMv5 conversions are under way now, with OMAP1 being the only one still not in linux-next yet, and PXA completing the set.
Most of the patches are unchanged from before, furtunately the PXA code is fairly stable. I addressed Robert's comments, pulled in two patches from Dmitry, and added the last a the final four patches to finish off the multiplatform conversion.
I hope someone is left to test these on PXA: if this works, I'd like to merge it for 5.19. A git tree with these is avaialable for testing at
https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc.git/log/?h=pxa-multi...
Unfortunately that crashes for me when trying to boot from ide. Bisect points to the last patch of the series.
Thanks a lot for testing and the perfect bug report!
[ 1.403715] 8<--- cut here --- [ 1.403848] Unable to handle kernel paging request at virtual address feeb000e [ 1.404097] [feeb000e] *pgd=00000000
Ok, this is the PCI I/O space area, which starts at 0xfee00000, clearly the way I/O space gets mapped changed here. I don't yet see what happened, but it should be straightforward to find from here.
[ 1.416643] pcmcia_init_one from pcmcia_device_probe+0xe4/0x2a0 [ 1.416882] pcmcia_device_probe from really_probe+0xc8/0x3b4 [ 1.417070] really_probe from __driver_probe_device+0x9c/0x214 [ 1.417255] __driver_probe_device from driver_probe_device+0x38/0xe0 [ 1.417454] driver_probe_device from __device_attach_driver+0xa4/0x11c [ 1.417657] __device_attach_driver from bus_for_each_drv+0x88/0xd8 [ 1.417864] bus_for_each_drv from __device_attach+0xf4/0x194 [ 1.418047] __device_attach from bus_probe_device+0x8c/0x94 [ 1.418224] bus_probe_device from device_add+0x3d0/0x894 [ 1.418395] device_add from pcmcia_device_add+0x2ec/0x3e0 [ 1.418568] pcmcia_device_add from pcmcia_card_add+0xd4/0x1a0 [ 1.418756] pcmcia_card_add from pcmcia_bus_add+0x44/0x4c [ 1.418930] pcmcia_bus_add from socket_insert+0x12c/0x150 [ 1.419103] socket_insert from pccardd+0x398/0x44c [ 1.419257] pccardd from kthread+0xdc/0x114 [ 1.419400] kthread from ret_from_fork+0x14/0x2c [ 1.419569] Exception stack(0xc48a5fb0 to 0xc48a5ff8) [ 1.419735] 5fa0: 00000000 00000000 00000000 00000000 [ 1.419979] 5fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 [ 1.420222] 5fe0: 00000000 00000000 00000000 00000000 00000013 00000000 [ 1.420501] Code: 13570000 e1a06000 0a000043 e3a03002 (e5c03000) [ 1.420874] ---[ end trace 0000000000000000 ]---
# bad: [7643a9ca9f8e08f71e15f89dd74863635e981e03] ARM: pxa: convert to multiplatform # good: [3123109284176b1532874591f7c81f3837bbdc17] Linux 5.18-rc1 git bisect start 'HEAD' 'v5.18-rc1' # good: [9b03d7f95bd4d97101ecb8ea1e822103b81fdb2d] ARM: pxa: mainstone-wm97xx: use gpio lookup table git bisect good 9b03d7f95bd4d97101ecb8ea1e822103b81fdb2d # good: [764063eee7620ea9abb940068a7ad0e7f9efa1b6] cpufreq: pxa3: move clk register access to clk driver git bisect good 764063eee7620ea9abb940068a7ad0e7f9efa1b6 # good: [5153474f0a4388b7ddb59add4be73bfb42b2007f] ARM: mmp: remove tavorevb board support git bisect good 5153474f0a4388b7ddb59add4be73bfb42b2007f # good: [2746f7c78b428c8b01b691a29a972c08101ae343] ARM: PXA: fix multi-cpu build of xsc3 git bisect good 2746f7c78b428c8b01b691a29a972c08101ae343 # good: [73d5106e9489464eac84362705e93bcf3b376123] ARM: pxa: remove support for MTD_XIP git bisect good 73d5106e9489464eac84362705e93bcf3b376123 # first bad commit: [7643a9ca9f8e08f71e15f89dd74863635e981e03] ARM: pxa: convert to multiplatform
I'll back out this patch for now while investigating further.
Which machine did you hit this on? Is this on hardware or in qemu?
qemu, as always. borzoi, spitz, terrier, tosa, z2, and sx1 fail. Also, I just noticed that the failure is not always the same. z2 fails to boot from initrd, and sx1 fails to boot completely. I'll do another round of bisects.
Guenter
Arnd