On Mon, Jun 01, 2020 at 08:45:19PM -0700, Guenter Roeck wrote:
On 6/1/20 8:30 PM, Sasha Levin wrote:
On Mon, Jun 01, 2020 at 12:22:54PM -0700, Guenter Roeck wrote:
Anyway, my script also tells me:
Upstream commit a33a5d2d16cb ("genirq/generic_pending: Do not lose pending affinity update") upstream: v4.18-rc1 Fixes: 98229aa36caa ("x86/irq: Plug vector cleanup race") in linux-4.4.y: 996c591227d9 upstream: v4.5-rc2 Affected branches: linux-4.4.y (queued) linux-4.9.y (queued) linux-4.14.y
and, indeed, it looks like a33a5d2d16cb is missing in v4.14.y-queue.
I think that Greg's script didn't like a33a5d2d16cb pointing to the wrong "fixes:" commit - 996c591227d9 rather than 98229aa36caa.
Interesting. Makes me wonder how my script found the correct reference.
FWIW, my scripts always try translating any commit id pointed to by a tag into an "upstream commit id". This is mostly helpful with tags pointing to linux-next commits or commits in private trees, but I suppose it's also useful for cases such as the above :)
But then why did his script pick it up for 4.4.y and 4.9.y ?
I can try and guess: Greg's to-do list has an item named "Figure out why I had to apply a33a5d2d16cb manually" :)