On 03/10/2024 16:33, Moon Yeounsu wrote: > On 03/10/2024 15:19, Eric Dumazet wrote: >> It also does not know about conditional locking, it is quite useless. > So... What do you think about who wants to send the patch to silence > the Sparse's warning message, nevertheless? Fwiw, my experience is that if I can't explain the locking to sparse that's usually a sign that the code is too complex and needs to be rewritten. In general I'm in favour of patches to fix sparse warnings. In this case it looks like what's needed is __cond_acquires, but the patch to implement that in sparse[1] doesn't seem to have gotten anywhere near Luc's tree. (Yet it's present and occasionally used in the kernel...) CCing the sparse ML to find out why. -ed [1]: https://lore.kernel.org/all/CAHk-=wjZfO9hGqJ2_hGQG3U_XzSh9_XaXze=HgPdvJbgrvASfA@xxxxxxxxxxxxxx/#t
