Hi, Greg,
On Fri, 16 Oct 2020 at 10:52, Greg KH gregkh@linuxfoundation.org wrote:
But that's not what that commit says should happen, it says it fixes a commit that showed up in the 5.5 release.
Is it incorrect? Have you tested this on 5.4.y?
Well, this took quite a lot of digging, sorry for the delay (I'm definitely not a timers/vDSO expert, I was just trying to understand the reason for a strange Musl workaround [1]), but you're completely right. Only 5.5+ is affected, since it was at that time the clock_gettime64 entry point was introduced for ARM [2]. Sorry for the noise!
Thanks, Rui
[1] https://git.musl-libc.org/cgit/musl/commit/?id=4486c579cbf0d989080705f515d08... [2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/ar...