Memory Access issue with TX1 in Jetpack 4.3

Autonomous Machines Jetson & Embedded Systems Jetson TX1
,
#1

Hi Sir,

We upgraded the TX1 from Jetpack 4.2 to Jetpack 4.3 and found the Memory Access issue.
In the meanwhile, we checked the NV forum and found there is a patch file Nvidia released for the TK1. We would like to check if there is a file for the TX1 to solve this kind of issue.
Below are our test steps and log file.

  1. Upgraded TX1 from Jetpack 4.2 to Jetpack 4.3
  2. Tested with our application, there were random system crash issues that happened during the testing process.
  3. Tested again but disabled our application. No system crash issues happened.
  4. Re-verified in disabling our application but with a Memory R/W testing program. System crashes again in the testing.
  5. We searched the NV forum and found there is a patch file for the TK1.

Happens within 2 hours and up on some units and 7-10 hours on others.
Fault looks like this.

1[26512.079272] Unable to handle kernel write to read-only memory at virtual address 00000008
2[26512.091888] Mem abort info:
3[26512.098550] ESR = 0x96000045
4[26512.101659] Exception class = DABT (current EL), IL = 32 bits
5[26512.107697] SET = 0, FnV = 0
6[26512.110843] EA = 0, S1PTW = 0
7[26512.114154] Data abort info:
8[26512.117349] ISV = 0, ISS = 0x00000045
9[26512.123082] CM = 0, WnR = 1
10[26512.126167] user pgtable: 4k pages, 39-bit VAs, pgd = ffffffc0cbde6000
11[26512.132900] [0000000000000008] *pgd=0000000000000000, *pud=0000000000000000
12[26512.140231] Internal error: Oops: 96000045 [#1] PREEMPT SMP
13[26512.145791] Modules linked in: fuse zram bcmdhd cfg80211 spidev nvgpu bluedroid_pm ip_tables x_tables
14[26512.155069] CPU: 3 PID: 12414 Comm: systemctl Not tainted 4.9.140-tegra #1
15[26512.161928] Hardware name: Jetson-TX1 (DT)
16[26512.166013] task: ffffffc03a9d9c00 task.stack: ffffffc0858cc000
17[26512.171925] PC is at unlink_anon_vmas+0xcc/0x230
18[26512.176531] LR is at unlink_anon_vmas+0xb8/0x230
19[26512.181136] pc : [<ffffff8008210cf4>] lr : [<ffffff8008210ce0>] pstate: 60400045
20[26512.188513] sp : ffffffc0858cfc80
21[26512.191817] x29: ffffffc0858cfc80 x28: ffffffc0378552d0
22[26512.197129] x27: 0000000000000001 x26: ffffff800a0831d4
23[26512.202441] x25: dead000000000100 x24: dead000000000200
24[26512.207751] x23: ffffffc039994580 x22: ffffffc0378552d0
25[26512.213064] x21: ffffffc0399945f8 x20: fffffffffffffff0
26[26512.218377] x19: ffffffc005e80040 x18: 00000000000005c3
27[26512.223688] x17: ffffffff00000001 x16: 0000000000000000
28[26512.229000] x15: 0000007fa931de08 x14: 0000007fa932b308
29[26512.234311] x13: 0000000000000000 x12: 0000000000000000
30[26512.239622] x11: ffffffc03a245880 x10: 00000055b9601000
31[26512.244935] x9 : ffffffc08ea13108 x8 : 0000000000000000
32[26512.250246] x7 : ffffffc005e80060 x6 : 0000007fa8e62000
33[26512.255556] x5 : 0000000000000000 x4 : ffffff800a167000
34[26512.260866] x3 : 0000000000000000 x2 : ffffffc0399945f8
35[26512.266175] x1 : ffffffc005e80040 x0 : ffffffc005e802e0
36[26512.271484]
37[26512.272970] Process systemctl (pid: 12414, stack limit = 0xffffffc0858cc000)
38[26512.280000] Call trace:
39[26512.282441] [<ffffff8008210cf4>] unlink_anon_vmas+0xcc/0x230
40[26512.288087] [<ffffff800820001c>] free_pgtables+0xcc/0x130
41[26512.293472] [<ffffff800820a8d4>] exit_mmap+0xac/0x118
42[26512.298512] [<ffffff80080b0548>] mmput+0x60/0x130
43[26512.303203] [<ffffff80080b961c>] do_exit+0x26c/0xa08
44[26512.308155] [<ffffff80080b9e48>] do_group_exit+0x40/0xa8
45[26512.313453] [<ffffff80080b9ed0>] __wake_up_parent+0x0/0x40
46[26512.318926] [<ffffff80080838c0>] el0_svc_naked+0x34/0x38
47[26512.324226] ---[ end trace f8476a12bd2f1ecd ]---
48[26512.337619] Fixing recursive fault but reboot is needed!
#2

Suggest to move on the latest JetPack 4.6.2 for better support. Thanks

#3

Hi Sir,

This upgraded request is from our end customer’s application upgrade.
Since the TX1 system is still deployed and used in the field site, this patch file is needed to release our customer’s pain point.

Thank you!!

Frank

#4

Which patch it is? Please help to ponit it out, then can double check if existing or can be back posting.

1 Like
#5

We refer to the below link from the NV forum the TK1 patch to resolve the 4GB DRAM memory access issue.

#6

Hi Sir,

Do you have similar patch files like TK1 that we can use on the TX1?
We followed your suggestion and upgraded the TX1 system to the Jetpack 4.6.2, the same, we face the Memory access issue in this version. (Refer to the attached log file.)
Do you have any solution for this issue?
Thanks!

serialLogs_test19_jetpack462.txt (27.5 KB)

#8

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

Hi,
The patch is not required for TX1. This should be a different case. For further check, please share steps so that we can replicate the issue on TX1 developer kit and do investigation.

#10