mbox series

[5.10,0/1] mt76: move mt76_init_tx_queue in common code

Message ID 20230112115850.9208-1-n.zhandarovich@fintech.ru
Headers show
Series mt76: move mt76_init_tx_queue in common code | expand

Message

Nikita Zhandarovich Jan. 12, 2023, 11:58 a.m. UTC 
Svace has identified unchecked return value of mt7615_init_tx_queue
function in 5.10 branch, even though it makes sense to track it
instead. This issue is fixed in upstream version by Lorenzo's patch.

The same patch can be cleanly applied to the 5.10 branch.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Comments

Greg Kroah-Hartman Jan. 12, 2023, 12:58 p.m. UTC | #1 
On Thu, Jan 12, 2023 at 03:58:49AM -0800, Nikita Zhandarovich wrote:
> Svace has identified unchecked return value of mt7615_init_tx_queue
> function in 5.10 branch, even though it makes sense to track it
> instead. This issue is fixed in upstream version by Lorenzo's patch.
> 
> The same patch can be cleanly applied to the 5.10 branch.

I do not understand, what issue/bug does this fix?  And how can you
trigger it?  And why only worry about the 5.10.y kernel branch?

thanks,

greg k-h
Nikita Zhandarovich Jan. 13, 2023, 3:04 p.m. UTC | #2 
My apologies, I should've have explained my reasoning better.

1. My issue with 5.10 version of mt7615_init_tx_queues() in drivers/net/wireless/mediatek/mt76/mt7615/dma.c is that return value of final call to mt7615_init_tx_queue() is not taken into account
when returning result of mt7615_init_tx_queues(). So, if last mt7615_init_tx_queue() fails (due to memory issues, for instance), parent function will still erroneously return 0.

2. To correct the issue, I turned to Lorenzo's patch in b671da33d1c5973f90f098ff66a91953691df582 which solves my petit problem as well as rewrites a single mt76_init_tx_queue() function to be used
across all mt76 drivers.

3. I was torn between writing my own little patch to fix a single mistake or use an existing one that increases code readability and uniformity. I settled on latter.

4. As for this patch exclusivity to 5.10.y branch, I have an incentive to prioritise prioritize 5.10. Wasn't sure I should be the one to suggest the patch for other branches.

Thanks,

Nikita
Greg Kroah-Hartman Jan. 14, 2023, 1:46 p.m. UTC | #3 
On Fri, Jan 13, 2023 at 07:04:45AM -0800, Nikita Zhandarovich wrote:
> My apologies, I should've have explained my reasoning better.

Reasoning for what?

Sorry, I have no context here, please properly quote emails so that we
have a hint as to what is going on.  Remember, some of us get 1000+ a
day that we need to process somehow...

thanks,

greg k-h