RE: [PATCH v3] seccomp: Improve performace by optimizing rmb()

> > On Feb 24, 2021, at 12:03 AM, wanghongzhe > wrote: > > > > As Kees haved accepted the v2 patch at a381b70a1 which just replaced > > rmb() with smp_rmb(), this patch will base on that and just adjust the > > smp_rmb() to the correct position. > > > > As the original comment shown (and indeed i

RE: [PATCH] seccomp: Improve performance by optimizing memory barrier

> From: Leon Romanovsky [mailto:l...@kernel.org] > Sent: Monday, February 8, 2021 2:44 PM > To: Wanghongzhe (Hongzhe, EulerOS) > Cc: keesc...@chromium.org; l...@amacapital.net; w...@chromium.org; > a...@kernel.org; dan...@iogearbox.net; and...@kernel.org; ka...@fb.com; > songli

RE: [PATCH v1 1/1] Firstly, as Andy mentioned, this should be smp_rmb() instead of rmb(). considering that TSYNC is a cross-thread situation, and rmb() is a mandatory barrier which should not be used

> Let's start with a patch that just replaces rmb() with smp_rmb() and then work > on optimizing. Can you provide performance numbers that show > rmb() (and soon smp_rmb()) is causing actual problems here? Ok, I will send a patch that just replaces rmb() with smp_rmb() and give performance numbers

RE: [PATCH] seccomp: Improve performance by optimizing memory barrier

>> On Feb 1, 2021, at 4:06 AM, wanghongzhe wrote: >> >> If a thread(A)'s TSYNC flag is set from seccomp(), then it will >> synchronize its seccomp filter to other threads(B) in same thread >> group. To avoid race condition, seccomp puts rmb() between reading the >> mode and filter in seccomp