On Thu, Mar 25, 2021 at 04:56:18PM +0800, Li Wang wrote: > Hi Heiko, > > On Wed, Mar 24, 2021 at 5:58 AM Heiko Carstens <hca@xxxxxxxxxxxxx> wrote: > > > Li Wang reported that clock_gettime(CLOCK_MONOTONIC_RAW, ...) does not > > work correctly on s390 via vdso. Debugging this also revealed an > > unrelated bug (first patch). > > > > The second patch fixes the problem: the tod clock steering parameters > > required by __arch_get_hw_counter() are only present within the first > > element of the _vdso_data array and not at all within the _timens_data > > array. > > > > Instead of working around this simply provide an s390 specific vdso > > data page which contains the tod clock steering parameters. > > > > This allows also to remove ARCH_HAS_VDSO_DATA again. > > > > Heiko Carstens (3): > > s390/vdso: fix tod clock steering > > s390/vdso: fix arch_data access for __arch_get_hw_counter() > > lib/vdso: remove struct arch_vdso_data from vdso data struct > > > > Thanks for the quick fix! I confirmed these patches work for me. > (tested with latest mainline kernel v5.12-rc4) > > Tested-by: Li Wang <liwang@xxxxxxxxxx> Thanks a lot for confirming! However I decided to go with the simple variant: https://lore.kernel.org/linux-s390/YFnxr1ZlMIOIqjfq@osiris/T/#m26f94fd8ac048421a4a8870e7259a09f97840a3e May I add your Tested-by there as well?
