Skip to contentSkip to conversation

Panic in simple_map_write called from cfi_probe_chip

DISCUSSION Eric Sesterhenn 2 participants 2 messages Jun 14, 2006

lore
Share & tools
lore
#0 lore raw

hi,

I decided to play around with all the new shiny stuff in 2.6.17-rc6-mm2, and tried to build a nearly allyesconfig kernel. After some tries, I got to the point where it still runs after uncompressing, and got into a panic.

[   31.270647] Photron PNC-2000 flash mapping: 400000 at bf000000
[   31.270737] BUG: unable to handle kernel paging request at virtual address bf000000
[   31.270869]  printing eip:
[   31.270925] c09d516e
[   31.270996] *pde = 00000000
[   31.271051] Oops: 0002 [#1]
[   31.271120] 4K_STACKS PREEMPT
[   31.271270] last sysfs file:
[   31.271340] Modules linked in:
[   31.271453] CPU:    0
[   31.271454] EIP:    0060:[<c09d516e>]    Not tainted VLI
[   31.271456] EFLAGS: 00010286   (2.6.17-rc6-mm2 #7)
[   31.271652] EIP is at simple_map_write+0x4f/0x76
[   31.271737] eax: f0f0f0f0   ebx: c12ed720   ecx: 00000000   edx: 00000020
[   31.271798] esi: dffd19e0   edi: bf000000   ebp: dffd1a0c   esp: dffd19e0
[   31.271885] ds: 007b   es: 007b   ss: 0068
[   31.271941] Process idle (pid: 1, threadinfo=dffd1000 task=dff86b40)
[   31.272032] Stack: f0f0f0f0 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[   31.272544]        dffd1dd8 c09d511f dffd1dd8 dffd1f04 c09be78e f0f0f0f0 00000000 00000000
[   31.273055]        00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000004
[   31.273563] Call Trace:
[   31.277555]  [<c09be78e>] cfi_probe_chip+0x17a/0x13b0
[   31.281571]  [<c09d11c0>] mtd_do_chip_probe+0x73/0x323
[   31.285518]  [<c09be612>] cfi_probe+0xd/0xf
[   31.289447]  [<c09be1a3>] do_map_probe+0x3b/0xbc
[   31.293376]  [<c190e6ab>] init_pnc2000+0x30/0x60
[   31.297311]  [<c01002a1>] init+0x81/0x1df
[   31.297476]  [<c0101005>] kernel_thread_helper+0x5/0xb
[   31.303646] Code: f8 01 75 0b 0f b6 45 d4 03 7b 0c 88 07 eb 35 83 f8 02 75 0c 0f b7 45 d4 03 
7b 0c 66 89 07 eb 24 83 f8 04 75 0a 03 7b 0c 8b 45 d4 <89> 07 eb 15 7e 13 89 c1 03 7b 0c c1 e9
02 f3 a5 89 c1 83 e1 03
[   31.306933] EIP: [<c09d516e>] simple_map_write+0x4f/0x76 SS:ESP 0068:dffd19e0
[   31.307120]  <0>Kernel panic - not syncing: Attempted to kill init!
[   31.307241]

it seems the offendling line is in include/linux/map.h where inline_map_write() calls __raw_writel(datum.x[0], map->virt + ofs);

due to some inlines not all calls show up here.. basically cfi_probe_chip() calls cfi_send_gen_cmd() in cfi.h which calls map_write() which in turn calls inline_map_write(). Since eax is filles with 0xF0 i guess it is the first call to cfi_send_gen_cmd() in cfi_probe_chip() cfi_send_gen_cmd(0xF0, 0, base, map, cfi, cfi->device_type, NULL);

Greetings, Eric

#1 lore raw
Quoted 7 lines · Eric Sesterhenn

I decided to play around with all the new shiny stuff in 2.6.17-rc6-mm2, and tried to build a nearly allyesconfig kernel. After some tries, I got to the point where it still runs after uncompressing, and got into a panic.

[   31.270647] Photron PNC-2000 flash mapping: 400000 at bf000000
[   31.270737] BUG: unable to handle kernel paging request at virtual address bf000000 

Yeah. The Photron PNC-2000 mapping assumes that the flash was already mapped. I'm not entirely sure why it does that. I'm also vaguely surprised that it's an x86 device -- for some reason I thought it was MIPS, which would have made bf000000 make a lot more sense.

Signature · 1 line
dwmw2