lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  PHC 
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date:   Tue, 28 May 2019 02:52:57 +0000
From:   Chris Packham <>
To:     "" <>,
        Paul Burton <>,
        James Hogan <>
CC:     "" <>,
        "" <>,
        Hamish Martin <>
Subject: MIPS r4k cache operations with SMP enabled


I'm trying to port a fairly old Broadcom integrated chip (BCM6818) to 
the latest Linux kernel using the mips/bmips support.

The chip has a BMIPS4355 core. This has two "thread processors" (cpu 
cores) with separate I-caches but a shared D-cache.

I've got things booting but I encounter the following BUG()

BUG: using smp_processor_id() in preemptible [00000000] code: swapper/0/1
caller is blast_dcache16+0x24/0x154
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.1.0-at1 #5
Stack : 00000036 8008d0d0 806a0000 807c0000 80754e10 0000000b 80754684 
         80900000 8f828424 807986e7 8071348c 00000000 10008f00 8f831c30 
         00000000 00000000 80920000 00000056 00002335 00000000 807a0000 
         6d6d3a20 00000000 00000056 73776170 00000000 ffffffff 10008f01 
         80790000 00002cc2 ffffffff 80900000 00000010 8f83198c 00000000 
Call Trace:
[<8001c208>] show_stack+0x30/0x100
[<8063282c>] dump_stack+0x9c/0xd0
[<802f1cec>] debug_smp_processor_id+0xfc/0x110
[<8002e274>] blast_dcache16+0x24/0x154
[<80122978>] map_vm_area+0x58/0x70
[<80123888>] __vmalloc_node_range+0x1fc/0x2b4
[<80123b54>] vmalloc+0x44/0x50
[<807d15d0>] jffs2_zlib_init+0x24/0x94
[<807d1354>] jffs2_compressors_init+0x10/0x30
[<807d151c>] init_jffs2_fs+0x68/0xf8
[<8001016c>] do_one_initcall+0x7c/0x1f0
[<807bee30>] kernel_init_freeable+0x17c/0x258
[<80650d1c>] kernel_init+0x10/0xf8
[<80015e6c>] ret_from_kernel_thread+0x14/0x1c

In blast_dcache16 current_cpu_data is used which invokes 
smp_processor_id() triggering the BUG(). I can fix this by sprinkling 
preempt_disable/preempt_enable through arch/mips/mm/c-r4k.c but that 
seems kind of wrong. Does anyone have any suggestion as to the right way 
to avoid this BUG()?


Powered by blists - more mailing lists