On Wed, Jan 15, 2025 at 7:01 PM Michal Koutný <mkoutny@xxxxxxxx> wrote: > > On Wed, Jan 15, 2025 at 01:05:21PM +0800, Jin Guojie <guojie.jin@xxxxxxxxx> wrote: > > V2: > > * call fmeter_init() when writing 0 to the memory_pressure_enabled > > Thanks for taking into account the feedback. > > I'm still curious -- is this: > a) a new LTP test, > b) a new failure, > c) an old failure, new look > or anything else? For the three situations a, b, and c you mentioned, none of them is true. In fact, this "case cpuset_memory_pressure" has been in LTP for a long time, and the same error will occur when running in multiple kernel versions. The method to reproduce the error is provided in the description of patch v1. https://lore.kernel.org/cgroups/CA+B+MYRNsdKcYxC8kbyzVrdH9fT8c2if5UxGguKep36ZHe6HMQ@xxxxxxxxxxxxxx/T/#u The fmeter_getrate() function in the kernel has not been modified for a long time, so currently LTP "case cpuset_memory_pressure" still produces the same error. > > Michal
