32-bit powerpc, aty128fb: vmap allocation for size 135168 failed

Started by Meelis Roos on Aug 18, 2017 · 5 messages from 2 authors · Last reply Aug 22, 2017

open the first message on its own page

Message details and related content
I was trying 4.13.0-rc5-00075-gac9a40905a61 on my PowerMac G4 with 1G 
RAM and after some time of sddm respawning and X trying to restart, 
dmesg is full of messages about vmap allocation failures.

Maybe the aty128fb is leaking ROM allocations or something like that?

sddm has been crashing eearlier too but I have not investigated it yet, 
but right after reboot the messages are about ATI ROM contents:

Aug 17 23:53:57 pohl kernel: [ 2940.146546] aty128fb 0000:00:10.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x1111
Aug 17 23:54:02 pohl kernel: [ 2944.804838] aty128fb 0000:00:10.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x1110
Aug 17 23:54:06 pohl kernel: [ 2948.992457] sddm[14039]: unhandled signal 11 at 00000030 nip 00000030 lr 0f55f858 code 30001

Then it changes to groups like this:
Aug 17 23:54:29 pohl kernel: [ 2971.514484] sddm[14093]: unhandled signal 11 at 00000090 nip 00000090 lr 0f55f858 code 30001
Aug 17 23:54:30 pohl kernel: [ 2972.994486] aty128fb 0000:00:10.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x1111
Aug 17 23:54:30 pohl kernel: [ 2973.040595] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size
Aug 17 23:54:33 pohl kernel: [ 2976.245220] aty128fb 0000:00:10.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x1111
Aug 17 23:54:33 pohl kernel: [ 2976.295452] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size

And finally it becomes just vmalloc errors, no ATI ROM messages at all:

[32075.316981] sddm[14563]: unhandled signal 11 at 00000050 nip 00000050 lr 0f55f858 code 30001
[32076.766965] vmap allocation for size 135168 failed: use vmalloc=<size> to increase size
[32076.788476] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size
[32079.124735] vmap allocation for size 135168 failed: use vmalloc=<size> to increase size
[32079.146326] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size
[32081.305352] sddm[14590]: unhandled signal 11 at 00000050 nip 00000050 lr 0f55f858 code 30001
[32082.768060] vmap allocation for size 135168 failed: use vmalloc=<size> to increase size
[32082.789530] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size
[32085.125847] vmap allocation for size 135168 failed: use vmalloc=<size> to increase size
[32085.147228] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size
[32087.311193] sddm[14617]: unhandled signal 11 at 00430068 nip 00430068 lr 0f55f858 code 30001
[32088.767983] vmap allocation for size 135168 failed: use vmalloc=<size> to increase size
[32088.789536] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size
[32091.108732] vmap allocation for size 135168 failed: use vmalloc=<size> to increase size
[32091.130348] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size
[32093.285222] sddm[14644]: unhandled signal 11 at 00000050 nip 00000050 lr 0f55f858 code 30001
[32094.767678] vmap allocation for size 135168 failed: use vmalloc=<size> to increase size
[32094.789329] vmap allocation for size 3149824 failed: use vmalloc=<size> to increase size

53 more lines

Message details and related content
Meelis Roos [off-list ref] writes:
I was trying 4.13.0-rc5-00075-gac9a40905a61 on my PowerMac G4 with 1G 
RAM and after some time of sddm respawning and X trying to restart, 
dmesg is full of messages about vmap allocation failures.
Did it just start happening? ie. did rc4 work?

cheers
Message details and related content
quoted
I was trying 4.13.0-rc5-00075-gac9a40905a61 on my PowerMac G4 with 1G 
RAM and after some time of sddm respawning and X trying to restart, 
dmesg is full of messages about vmap allocation failures.
Did it just start happening? ie. did rc4 work?
No, rc4 was the first one I tried after 4.12 and it already had the 
problem. Not sure about 4.12 to be honest, will soee some day.

-- 
Meelis Roos (mroos@linux.ee)
Message details and related content
Meelis Roos [off-list ref] writes:
quoted
I was trying 4.13.0-rc5-00075-gac9a40905a61 on my PowerMac G4 with 1G 
RAM and after some time of sddm respawning and X trying to restart, 
dmesg is full of messages about vmap allocation failures.
Did it just start happening? ie. did rc4 work?
It goes back to at least 4.0 - that's the oldest kernel I had laying 
around precompiled. The messages about ROM signature changed somewehere 
between 4.0 and 4.7 (4.7 is already like 4.13) but after some time, the 
same vmalloc errors appear.

Maybe the userspace has changed with more respawning that brings the 
problem out.

I tried to read the code but I do not understand it yet. The warning 
seems to come from generic pci_map_rom() checking ROM size, and 
returning rom pointer to aty128fb (it returns resource size too but that 
is ignored). aty128fb starts to look at the x86 PCI ROM signature again 
but does not tell that the signature is missing. How come?

-- 
Meelis Roos (mroos@linux.ee)
Message details and related content
Meelis Roos [off-list ref] writes:
quoted
Meelis Roos [off-list ref] writes:
quoted
I was trying 4.13.0-rc5-00075-gac9a40905a61 on my PowerMac G4 with 1G 
RAM and after some time of sddm respawning and X trying to restart, 
dmesg is full of messages about vmap allocation failures.
Did it just start happening? ie. did rc4 work?
It goes back to at least 4.0 - that's the oldest kernel I had laying 
around precompiled. The messages about ROM signature changed somewehere 
between 4.0 and 4.7 (4.7 is already like 4.13) but after some time, the 
same vmalloc errors appear.

Maybe the userspace has changed with more respawning that brings the 
problem out.

I tried to read the code but I do not understand it yet. The warning 
seems to come from generic pci_map_rom() checking ROM size, and 
returning rom pointer to aty128fb (it returns resource size too but that 
is ignored). aty128fb starts to look at the x86 PCI ROM signature again 
but does not tell that the signature is missing. How come?
OK thanks. Unfortunately I can't really help on the graphics side of
things. Hopefully someone on Cc has more clues than me.

cheers
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help