Bug 218826 - Regression of e1000e (I219-LM) from 6.1.90 to 6.6.30
Summary: Regression of e1000e (I219-LM) from 6.1.90 to 6.6.30
Status: RESOLVED CODE_FIX
Alias: None
Product: Drivers
Classification: Unclassified
Component: Network (show other bugs)
Hardware: All Linux
: P3 normal
Assignee: drivers_network@kernel-bugs.osdl.org
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2024-05-10 14:42 UTC by lukas.probsthain
Modified: 2024-05-20 17:34 UTC (History)
3 users (show)

See Also:
Kernel Version:
Subsystem:
Regression: No
Bisected commit-id:


Attachments
Kernel log upon connecting the ethernet cable (6.11 KB, text/plain)
2024-05-10 14:42 UTC, lukas.probsthain
Details
lspci (9.41 KB, text/plain)
2024-05-10 14:42 UTC, lukas.probsthain
Details
Kernel log for 6.8.9-3 (3.29 KB, text/plain)
2024-05-11 08:32 UTC, lukas.probsthain
Details

Description lukas.probsthain 2024-05-10 14:42:06 UTC
Created attachment 306280 [details]
Kernel log upon connecting the ethernet cable

After updating from kernel version 6.1.90 to 6.6.30, the e1000e driver exhibits a regression on a Lenovo Thinkpad T480 with an Intel I219-LM Ethernet controller. The system experiences a freeze when an Ethernet cable is plugged in. The issue is not present in the previous kernel version 6.1.90.

System Information:
- Model: Lenovo Thinkpad T480
- BIOS Version: N24ET76W (1.51) dated 02/27/2024
- Ethernet Controller: Intel Corporation Ethernet Connection (4) I219-LM (rev 21)
- Kernel Module in Use: e1000e
- Operating System: Manjaro Linux, kernel version 6.6.30-1

Steps to Reproduce:
1. Boot system with kernel version 6.6.30.
2. Connect the Ethernet cable to the laptop.
3. Observe that the system freezes.

Expected Behavior:
The system should remain stable and maintain network connectivity without freezing when the Ethernet cable is connected.

Actual Behavior:
The system freezes immediately upon plugging in the Ethernet cable.

Additional Information:
The regression seems to be introduced in one of the updates between kernel versions 6.1.90 and 6.6.30. The issue does not occur with the older kernel version 6.1.90.
Comment 1 lukas.probsthain 2024-05-10 14:42:58 UTC
Created attachment 306281 [details]
lspci

Added device list
Comment 2 Artem S. Tashkinov 2024-05-11 07:52:44 UTC
Please try to reproduce in 6.8.9.

If it's still broken, please perform regression testing:

https://docs.kernel.org/admin-guide/bug-bisect.html
Comment 3 lukas.probsthain 2024-05-11 08:32:10 UTC
Created attachment 306282 [details]
Kernel log for 6.8.9-3

The problem seems to persist with 6.8.9

Next I'll build the kernel manually and search for the bad commit with git bisect.
Comment 4 Artem S. Tashkinov 2024-05-12 08:26:06 UTC
Performing git bisect between the two kernels so far out makes little sense.

I'm sure you can cut it down to two consecutive kernel releases, e.g. 6.1.0 and 6.2.0, and that way it will be far faster and easier to bisect.
Comment 5 Artem S. Tashkinov 2024-05-12 08:31:24 UTC
Also please try to apply this patch and see if it helps:

https://patchwork.ozlabs.org/project/intel-wired-lan/patch/20240417190320.3159360-1-vitaly.lifshits@intel.com/
Comment 6 lukas.probsthain 2024-05-13 20:59:13 UTC
So far I was only able to test the kernel version 6.8 (@ e8f897f4afef0031fe618a8e94127a0934896aba) and it seems to work so far. I'll try the other versions in between 6.1 and 6.8 in the next days. But at this point it might be possible that the error is introduced by the kernel version deployed by manjaro. I have compared the config for compiling the kernel and at first sight nothing obvious was off.

I'll report back as soon as I have new information.
Comment 7 Bagas Sanjaya 2024-05-14 04:33:11 UTC
(In reply to lukas.probsthain from comment #6)
> So far I was only able to test the kernel version 6.8 (@
> e8f897f4afef0031fe618a8e94127a0934896aba) and it seems to work so far. I'll
> try the other versions in between 6.1 and 6.8 in the next days. But at this
> point it might be possible that the error is introduced by the kernel
> version deployed by manjaro. I have compared the config for compiling the
> kernel and at first sight nothing obvious was off.
> 
> I'll report back as soon as I have new information.

You may want further limit amount of commits to test by passing driver path to
git-bisect(1), like:

```
$ git bisect start -- drivers/net/ethernet/intel/e1000e
$ git bisect good v6.1
$ git bisect bad v6.8
```
Comment 8 Artem S. Tashkinov 2024-05-20 17:34:47 UTC
Must be fixed in 6.8.10. If not, please reopen.

Note You need to log in before you can comment on or make changes to this bug.