Could external jtag debugger trap page fault instead of kernel handler?

[Hsia-Jun Li <Randy.Li@synaptics.com>]

Hello

I am trying to debug an illegal memory access problem for a kernel 
driver. The JTAG could trap the kernel function likes 
die_kernel_fault(). But it won't provide a context environment in the 
place where this bug happens.

I know such page fault is more like a software interrupter, is there a 
way that I could move my debugging context back?

Here is what I got from gdb connected to openocd.
(gdb) bt 

#0  __do_kernel_fault (addr=16045481047390945544, addr@entry=264, 
esr=2516582468, regs=0xffffff802937b940)
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/arch/arm64/mm/fault.c:372 

#1  0xffffffc00802c174 in do_bad_area (regs=<optimized out>, 
esr=<optimized out>, far=16045481047390945544)
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/arch/arm64/mm/fault.c:485 

#2  do_translation_fault (far=16045481047390945544, esr=<optimized out>, 
regs=<optimized out>)
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/arch/arm64/mm/fault.c:782 

#3  0xffffffc00802c2cc in do_mem_abort (far=16045481047390945544, 
esr=2516582468, regs=0xffffff802937b940)
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/arch/arm64/mm/fault.c:918 

#4  0xffffffc00889738c in skb_dump (level=0xdead000000000108 "", 
skb=0xffffff802937b940, full_pkt=112)
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/net/core/skbuff.c:834 

#5  0x0000000000000000 in ?? () 

Backtrace stopped: not enough registers or memory available to unwind 
further
(gdb) info threads
   Id   Target Id 
         Frame
   1    Thread 1 "vs640.cpu_a53.0" (Name: vs640.cpu_a53.0, state: 
debug-request) 0xffffffc008898038 in nf_l4proto_log_inv
alid (skb=0xffffffc0086391a0 <psci_enter_idle_state>, state=0xffffffff, 
protonum=0 '\000', fmt=0x258208 "")
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/net/netfilter/nf_conntrack_proto.c:62
* 2    Thread 2 "vs640.cpu_a53.1" (Name: vs640.cpu_a53.1, state: 
breakpoint)    __do_kernel_fault (
     addr=16045481047390945544, addr@entry=264, esr=2516582468, 
regs=0xffffff802937b940)
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/arch/arm64/mm/fault.c:372
   3    Thread 3 "vs640.cpu_a53.2" (Name: vs640.cpu_a53.2, state: 
debug-request) 0xffffffc008898038 in nf_l4proto_log_inv
alid (skb=0xffffffc0086391a0 <psci_enter_idle_state>, state=0xffffffff, 
protonum=0 '\000', fmt=0x9f2270 "")
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/net/netfilter/nf_conntrack_proto.c:62
   4    Thread 4 "vs640.cpu_a53.3" (Name: vs640.cpu_a53.3, state: 
debug-request) 0xffffffc008898038 in nf_l4proto_log_inv
alid (skb=0xffffffc0086391a0 <psci_enter_idle_state>, state=0xffffffff, 
protonum=0 '\000', fmt=0x20ded8 "")
     at 
/Codebase_s/ranl/workspace/vs680/kernel/linux-syna/net/netfilter/nf_conntrack_proto.c:62

And here is the issue I am trying to debug:
[ 5478.863161] Unable to handle kernel paging request at virtual address 
dead000000000108
[ 5478.871093] Mem abort info:
[ 5478.873886]   ESR = 0x96000044
[ 5478.876938]   EC = 0x25: DABT (current EL), IL = 32 bits
[ 5478.882249]   SET = 0, FnV = 0
[ 5478.885301]   EA = 0, S1PTW = 0
[ 5478.888440]   FSC = 0x04: level 0 translation fault
[ 5478.893318] Data abort info:
[ 5478.896195]   ISV = 0, ISS = 0x00000044
[ 5478.900029]   CM = 0, WnR = 1
[ 5478.902996] [dead000000000108] address between user and kernel 
address ranges
[ 5478.910129] Internal error: Oops: 96000044 [#1] PREEMPT SMP
[ 5478.915701] Modules linked in: vpu videobuf2_memops v4l2_mem2mem 
videobuf2_v4l2 videobuf2_common videodev mc synap(O)
  pvrsrvkm syna_drm cfg80211 rfkill amp_bm
[ 5478.929897] CPU: 0 PID: 6613 Comm: gst-launch-1.0 Tainted: G        W 
  O      5.15.62 #1
[ 5478.937988] Hardware name: Synaptics VS640 EVK (DT)
[ 5478.942864] pstate: 204000c9 (nzCv daIF +PAN -UAO -TCO -DIT -SSBS 
BTYPE=--)
[ 5478.949825] pc : v4l2_m2m_buf_remove+0x50/0x90 [v4l2_mem2mem]
[ 5478.955582] lr : v4l2_m2m_buf_remove+0x20/0x90 [v4l2_mem2mem]
[ 5478.961329] sp : ffffff8026c4fa90
[ 5478.964643] x29: ffffff8026c4fa90 x28: ffffff8026c201b0 x27: 
0000000000000000
[ 5478.971784] x26: 0000000000000000 x25: ffffff8026c4fd28 x24: 
ffffff8009774270
[ 5478.978921] x23: ffffff8025f4c100 x22: 0000000000000001 x21: 
ffffff8025e1f3a0
[ 5478.986059] x20: ffffff8025e6d7b0 x19: ffffff8025e1f028 x18: 
0000000000000000
[ 5478.993197] x17: 0000000000000000 x16: 0000000000000000 x15: 
00000055a1955a10
[ 5479.000335] x14: 0000000000000000 x13: 0000000000000000 x12: 
0000000000000000
[ 5479.007472] x11: 0000000000000000 x10: 0000000000000000 x9 : 
0000000000000000
[ 5479.014609] x8 : 0000000000000000 x7 : 0000000000000000 x6 : 
0000000000000005
[ 5479.021746] x5 : dead000000000100 x4 : dead000000000122 x3 : 
dead000000000100
[ 5479.028884] x2 : dead000000000122 x1 : 0000000000000000 x0 : 
ffffff8025e1f3a0
[ 5479.036021] Call trace:
[ 5479.038467]  v4l2_m2m_buf_remove+0x50/0x90 [v4l2_mem2mem]
[ 5479.043867]  vb2ops_vdec_stop_streaming+0x98/0x160 [vpu]
[ 5479.049184]  __vb2_queue_cancel+0x30/0x210 [videobuf2_common]
[ 5479.054936]  vb2_core_streamoff+0x20/0xb4 [videobuf2_common]
[ 5479.060597]  vb2_streamoff+0x18/0x64 [videobuf2_v4l2]
[ 5479.065654]  v4l2_m2m_streamoff+0x70/0x140 [v4l2_mem2mem]
[ 5479.071055]  v4l2_m2m_ioctl_streamoff+0x14/0x20 [v4l2_mem2mem]
[ 5479.076889]  v4l_streamoff+0x20/0x30 [videodev]
[ 5479.081455]  __video_do_ioctl+0x174/0x3d0 [videodev]

Sincerely
-- 
Hsia-Jun(Randy) Li