Possible gcc 4.8.5 bug about RELOC_HIDE marcro

Jia He hejianet at gmail.com
Wed Sep 20 23:31:55 PDT 2017


I tried to build kernel 4.14-rc1 on a arm64 server in distro centos 7.3. 
The gcc version is 4.8.5

It was built successfully but failed to boot with the call trace below:

===========call trace begin==============

[    8.993531] Unable to handle kernel NULL pointer dereference at 
virtual address 0000c4a0
[    9.000668] Mem abort info:
[    9.000669]   Exception class = DABT (current EL), IL = 32 bits
[    9.000670]   SET = 0, FnV = 0
[    9.000670]   EA = 0, S1PTW = 0
[    9.000671] Data abort info:
[    9.000671]   ISV = 0, ISS = 0x00000005
[    9.000672]   CM = 0, WnR = 0
[    9.000674] user pgtable: 64k pages, 48-bit VAs, pgd = ffff8017ddf79c00
[    9.000675] [000000000000c4a0] *pgd=0000000000000000, 
*pud=0000000000000000
[    9.000678] Internal error: Oops: 96000005 [#1] SMP
[    9.000679] Modules linked in: sdhci_acpi ixgbe(+) mdio xhci_plat_hcd 
at803x xhci_hcd ahci_platform libahci_platform qcom_emac libahci usbcore 
sdhci ipv6 crc_ccitt
[    9.000693] CPU: 1 PID: 1073 Comm: kworker/1:1 Not tainted 4.14.0-rc1+ #5
[    9.000693] Hardware name: To be filled by O.E.M. To be filled by 
O.E.M./To be filled by O.E.M., BIOS 5.13 12/12/2012
[    9.000701] Workqueue: events_power_efficient process_srcu
[    9.000703] task: ffff8017cd498c00 task.stack: ffff00001bbe0000
[    9.000704] PC is at process_srcu+0x50/0x4bc
[    9.000706] LR is at process_srcu+0x48/0x4bc
[    9.000707] pc : [<ffff00000813fc30>] lr : [<ffff00000813fc28>] 
pstate: 60400145
[    9.000707] sp : ffff00001bbefcf0
[    9.000708] x29: ffff00001bbefcf0 x28: ffff8017f952c800
[    9.000710] x27: ffff000009271000 x26: ffff000009484c88
[    9.000711] x25: 0000000000000000 x24: ffff000009b5aca0
[    9.000713] x23: ffff8017f9530f00 x22: ffff000009b5aca8
[    9.000715] x21: ffff8017f952c800 x20: ffff000009b5ac00
[    9.000716] x19: ffff000009b5a9d8 x18: 0000ffffdd61b6c0
[    9.000721] x17: 0000000000000000 x16: 0000000000000000
[    9.000722] x15: 0000000000000000 x14: 0000000000000000
[    9.000724] x13: 0000000000000000 x12: 0000000000000000
[    9.000725] x11: 0000000000000000 x10: 0000000000000c80
[    9.000727] x9 : ffff00001bbefd30 x8 : ffff8017cd4998e0
[    9.000729] x7 : 0000000000000000 x6 : 000000000ab89a36
[    9.000730] x5 : 000000000ab89a36 x4 : 000000000000079e
[    9.000732] x3 : ffff8017f952c820 x2 : 000000000000c4a0
[    9.000733] x1 : 0000000000000000 x0 : 0000000000000000
[    9.000735] Process kworker/1:1 (pid: 1073, stack limit = 
0xffff00001bbe0000)
[    9.000736] Call trace:
[    9.000738] Exception stack(0xffff00001bbefbb0 to 0xffff00001bbefcf0)
[    9.000739] fba0: 0000000000000000 0000000000000000
[    9.000741] fbc0: 000000000000c4a0 ffff8017f952c820 000000000000079e 
000000000ab89a36
[    9.000742] fbe0: 000000000ab89a36 0000000000000000 ffff8017cd4998e0 
ffff00001bbefd30
[    9.000743] fc00: 0000000000000c80 0000000000000000 0000000000000000 
0000000000000000
[    9.000745] fc20: 0000000000000000 0000000000000000 0000000000000000 
0000000000000000
[    9.000746] fc40: 0000ffffdd61b6c0 ffff000009b5a9d8 ffff000009b5ac00 
ffff8017f952c800
[    9.000747] fc60: ffff000009b5aca8 ffff8017f9530f00 ffff000009b5aca0 
0000000000000000
[    9.000749] fc80: ffff000009484c88 ffff000009271000 ffff8017f952c800 
ffff00001bbefcf0
[    9.000750] fca0: ffff00000813fc28 ffff00001bbefcf0 ffff00000813fc30 
0000000060400145
[    9.000751] fcc0: ffff00001bbefcd0 ffff000008ac88dc ffffffffffffffff 
ffff00000813fc28
[    9.000752] fce0: ffff00001bbefcf0 ffff00000813fc30
[    9.000754] [<ffff00000813fc30>] process_srcu+0x50/0x4bc
[    9.000757] [<ffff0000080eac64>] process_one_work+0x16c/0x380
[    9.000759] [<ffff0000080eaed8>] worker_thread+0x60/0x3d4
[    9.000760] [<ffff0000080f182c>] kthread+0x10c/0x138
[    9.000762] [<ffff000008084d00>] ret_from_fork+0x10/0x20
[    9.000764] Code: aa1403e0 94262327 d28c4a02 8b020042 (c8dffc40)
[    9.000786] ---[ end trace 27afa0bd722ea1ea ]---
[    9.000787] Kernel panic - not syncing: Fatal exception
[    9.000800] SMP: stopping secondary CPUs
[    9.003437] Kernel Offset: disabled
[    9.003438] CPU features: 0x060418
[    9.003439] Memory Limit: none
[    9.340761] ---[ end Kernel panic - not syncing: Fatal exception

===========call trace end==============

I tried to disassemble the code and found the related lines:

Dump of assembler code for function process_srcu:
    0xffff00000813c5c4 <+0>:     stp     x29, x30, [sp,#-160]!
    0xffff00000813c5c8 <+4>:     mov     x29, sp
    0xffff00000813c5cc <+8>:     stp     x19, x20, [sp,#16]
    0xffff00000813c5d0 <+12>:    stp     x21, x22, [sp,#32]
    0xffff00000813c5d4 <+16>:    stp     x23, x24, [sp,#48]
    0xffff00000813c5d8 <+20>:    stp     x25, x26, [sp,#64]
    0xffff00000813c5dc <+24>:    stp     x27, x28, [sp,#80]
    0xffff00000813c5e0 <+28>:    mov     x24, x0
    0xffff00000813c5e4 <+32>:    sub     x0, x0, #0x6, lsl #12
    0xffff00000813c5e8 <+36>:    sub     x1, x0, #0x2c8
    0xffff00000813c5ec <+40>:    add     x19, x1, #0x6, lsl #12
    0xffff00000813c5f0 <+44>:    str     x0, [x29,#144]
    0xffff00000813c5f4 <+48>:    mov     x0, x30
    0xffff00000813c5f8 <+52>:    str     x1, [x29,#152]
    0xffff00000813c5fc <+56>:    add     x20, x19, #0x228
    0xffff00000813c600 <+60>:    bl      0xffff000008090830 <_mcount>
    0xffff00000813c604 <+64>:    mov     x0, x20
    0xffff00000813c608 <+68>:    bl      0xffff000008aa8554 <mutex_lock>
    0xffff00000813c60c <+72>:    mov     x2, #0x6250                     
// #25168
    0xffff00000813c610 <+76>:    add     x2, x2, x2
    ------>0xffff00000813c614 <+80>:    ldar    x0, [x2]         <------ 
panic in this line
    0xffff00000813c618 <+84>:    and     w0, w0, #0x3
    0xffff00000813c61c <+88>:    cbz     w0, 0xffff00000813c678 
<process_srcu+180>
    0xffff00000813c620 <+92>:    ldr     x2, [x24,#-120]
    0xffff00000813c624 <+96>:    and     w2, w2, #0x3
    0xffff00000813c628 <+100>:   cmp     w2, #0x1
    0xffff00000813c62c <+104>:   b.eq    0xffff00000813c9ac 
<process_srcu+1000>
    0xffff00000813c630 <+108>:   ldr     x2, [x24,#-120]

seems the compiler doesn't work correctly, should it be some thing like

add     x2, x2, x25 ??

instead of

add     x2, x2, x2

Besides, I git bisect and find this *kernel* patch cause the compiler bug:

commit    c350c008297643dad3c395c2fd92230142da5cf6
srcu: Prevent sdp->srcu_gp_seq_needed counter wrap

In this bug, srcu uses a percpu ptr which will call RELOC_HIDE. After I 
remove

the RELOC_HIDE code, this bug disappearred.


This bug is not in latest gcc version


Cheers,

Justin(Jia He)




More information about the linux-arm-kernel mailing list