This article is more than 1 year old
Huawei dev flamed for 'useless' Linux kernel code contributions
Time-wasting commits called out as effort to burnish submission metrics
Updated Last week, Linux kernel contributor Qu Wenruo scolded another code donor, Zhen Lei, for wasting kernel maintainers' time with unnecessary patches.
In a post to Zhen Lei and the rest of the Linux kernel mailing list, Wenruo said he recently found a patch removing a debug out-of-memory error message from a selftest used by btrfs, a file system supported by the Linux kernel.
"It's nothing special, some small cleanup work from some kernel newbie," wrote Wenruo. "But the mail address makes me cautious, '@huawei.com'."
This is not the first time similar harmless "cleanup" patches have come from Huawei, said Wenruo, who observed those fixes were also "useless."
"This makes me wonder, what is really going on here," he wrote, noting that a quick search found a number of patches to "cleanup" out-of-memory error messages or to fix misspellings.
Wenruo's theory is that Zhen Lei submitted this inconsequential patch for Key Performance Indicator (KPI) credit – to do something that gets recognized by an employee performance measurement system as meaningful work.
Wenruo said it's fine for new developers and students to submit these sorts of patches, noting that he started his kernel contributions in the same way and that he hopes these small patches will lead to long term contributions.
I have already seen several maintainers arguing with you on such 'cleanups,' and you're always defending yourself to try to get those patches merged
"But what you guys are doing is really KPI grabbing," he said. "I have already seen several maintainers arguing with you on such 'cleanups,' and you're always defending yourself to try to get those patches merged."
"You're sending the patch representing your company, by doing this you're really just damaging the already broken reputation."
Wenruo's response may seem overly harsh – we asked a Linux maintainer whether the upbraid is consistent with the Linux Code of Conduct but haven't heard back. However, the Linux kernel community in April had a bad experience with researchers from the University of Minnesota who submitted "hypocrite commits" – shoddy code – to test whether maintainers were paying attention. The backlash underscored how much work volunteers contribute and how much they resent it when their time is wasted.
The Register asked Zhen Lei to comment but we've not heard back.
In an email to The Register, Wenruo said, "Some Chinese tech companies are really pushing too hard by assigning almost impossible KPI goals, I think that's the root cause."
"This pushes their employees to do things without using their common sense. And obviously toxic company culture like 996 (9am to 9pm, 6 days a week) and destructive competition."
"I hope there will be less and less incidents like this, but without addressing the root cause, it's just a time of problem to hit the next incident."
Wenruo said as far as he was concerned, the University of Minnesota affair didn't play into this.
"The Minnesota incident is at least harder to detect, as those 'fixes' look like real fixes at a quick glance, only when digging deeper it shows some problems," he said. "But in the Huawei incident, it's pretty simple, all the 'cleanups' are just generated from checkpatch.pl
script."
Wenrou said he found the author's reply annoying and wished the author just saved maintainers' time by making clear that the 'cleanup' commits are just warnings from checkpatch.pl
.
- Linux 5.13 hits rc5, isn’t yet calm, Linus Torvalds is only mildly perturbed
- Huawei names first tablets, phones to run its Android-in-disguise HarmonyOS 2
- Linus Torvalds tells kernel list poster to 'SHUT THE HELL UP' for saying COVID-19 vaccines create 'new humanoid race'
- It took 'over 80 different developers' to review and fix 'mess' made by students who sneaked bad code into Linux
Zhen Lei did respond, however, on the mailing list, defending past kernel contributions and promising that subsequent contributions to the Linux community will be more consequential.
And in response to that, Wenruo offered a list of projects that would be more helpful to tackle.
The involvement of Huawei, which released its own Linux distribution and has become an IT policy flashpoint in the US and Europe over concerns about supply chain integrity and government spying, has made this dispute a matter of interest among software developers in China.
On Chinese Q&A site Zhihu, an individual purporting to be a Red Hat engineer dissected the mailing list exchange. Insofar as the discussion can be deciphered via algorithmic translation, the person making the post argues that the Huawei shouldn't even figure into the discussion because commits of this sort and concern about them happens all the time. Those responding to the post argue that if a Google employee made a similar commit that person would be lionized.
We also contacted Huawei, one of the top two contributors in the Linux v5.10 release alongside Intel, to ask whether the telecom equipment titan counts commits for KPI assessment of employees. A spokesperson in the US said she'd look into it but cautioned that the need to communicate with global teams might preclude an immediate response. ®
Updated to add
In an email to The Register after this story was published, Linux maintainer David Sterba confirmed there is an issue in the kernel's community with inconsequential contributions.
“I’d say yes that there are contributions that are too trivial and come in large numbers,” he said. “I myself would not connect it to something like KPI. I wasn’t familiar with that term before, but patch count boosting happens from time to time and is no that hard to spot.”
Sterba said that typically what happens is that people are encouraged to find less trivial ways to contribute once they’re comfortable with the patch sending workflow.
“Counting patches is perhaps the simplest contribution metric and it’s not news that companies may be using just that to measure performance of employees or to show up high in the kernel contribution stats (e.g. periodically published by LWN),” he said.
Sterba expressed doubt that Wenruo’s admonition would run afoul of the Linux Code of Conduct.
“I don’t think it’s in conflict with CoC, I’d reserve that for serious issues,” he said. “For me it’s been on the technical grounds, but there have been some repetitive patterns that ask for feedback regarding the motivations and perhaps to be more explicit about the stance of (part of) the community.”
Interest in this particular exchange, he suggested, had little to do with the University of Minnesota incident and more to do with the contributor’s ties to Huawei.
“I don’t think much changed after that incident,” he said. “It caused some extra work, for me personally it was ‘just another patch’. This time it’s calling out some big company. The perception is different inside China and in the rest of the world, so this maybe makes it interesting. Other than mentioning the KPI and reputation, it’s the type of general advice that anybody can get.”