From bb7e161d5a06f1bbd5a98cf868eab356c6c1bac2 Mon Sep 17 00:00:00 2001 From: Jim Warner Date: Thu, 16 Mar 2023 00:00:00 -0500 Subject: [PATCH] top: address the missing 'guest' tics for summary area Well this is embarrassing. After repeatedly flogging a horse (represented by issue #274) I was certain it was dead. But, it turns out that the darn thing yet lived. In fact, the bug that was patched was not even the one the poster experienced. Now merge request #173 finally penetrated my foggy brain and explicated the real bug. Since forever (linux 2.6), top has ignored those guest and guest_nice fields in /proc/stat. When many virtual machines were running that overhead went unrecognized. So, this commit simply adds those tics to the 'system' figures so that it can be seen in text or graph modes. Reference(s): https://gitlab.com/procps-ng/procps/-/merge_requests/173 https://gitlab.com/procps-ng/procps/-/issues/274 . Mar 2023, avoid keystroke '%Cpu' distortions commit 7e33fc47c642aceea8ad53e86a6797239b91f5f9 Signed-off-by: Jim Warner --- man/top.1 | 7 +++++-- src/top/top.c | 7 +++++++ 2 files changed, 12 insertions(+), 2 deletions(-) diff --git a/man/top.1 b/man/top.1 index 5a77beb1..8bf804aa 100644 --- a/man/top.1 +++ b/man/top.1 @@ -527,6 +527,9 @@ Depending on your kernel version, the \fBst\fR field may not be shown. \fBst\fR : time stolen from this vm by the hypervisor .fi +The `sy' value above also reflects the time running a virtual \*(Pu +for guest operating systems, including those that have been niced. + Beyond the first tasks/threads line, there are alternate \*(PU display modes available via the 4-way `t' \*(CT. They show an abbreviated summary consisting of these elements: @@ -537,8 +540,8 @@ They show an abbreviated summary consisting of these elements: .fi Where: a) is the `user' (us + ni) percentage; b) is the `system' -(sy + hi + si) percentage; c) is the total percentage; and d) is -one of two visual graphs of those representations. +(sy + hi + si + guests) percentage; c) is the total percentage; +and d) is one of two visual graphs of those representations. Such graphs also reflect separate `user' and `system' portions. If the `4' \*(CT is used to yield more than two cpus per line, diff --git a/src/top/top.c b/src/top/top.c index 26a9adb5..02ef8f6d 100644 --- a/src/top/top.c +++ b/src/top/top.c @@ -286,6 +286,7 @@ static enum stat_item Stat_items[] = { STAT_TIC_DELTA_NICE, STAT_TIC_DELTA_IDLE, STAT_TIC_DELTA_IOWAIT, STAT_TIC_DELTA_IRQ, STAT_TIC_DELTA_SOFTIRQ, STAT_TIC_DELTA_STOLEN, + STAT_TIC_DELTA_GUEST, STAT_TIC_DELTA_GUEST_NICE, STAT_TIC_SUM_DELTA_USER, STAT_TIC_SUM_DELTA_SYSTEM, #ifdef CORE_TYPE_NO STAT_TIC_SUM_DELTA_TOTAL }; @@ -298,6 +299,7 @@ enum Rel_statitems { stat_NI, stat_IL, stat_IO, stat_IR, stat_SI, stat_ST, + stat_GU, stat_GN, stat_SUM_USR, stat_SUM_SYS, #ifdef CORE_TYPE_NO stat_SUM_TOT }; @@ -6340,6 +6342,11 @@ static int sum_tics (struct stat_stack *this, const char *pfx, int nobuf) { if (1 > tot_frme) idl_frme = tot_frme = 1; scale = 100.0 / (float)tot_frme; + /* account for VM tics not otherwise provided for ... + ( with xtra-procps-debug.h, can't use PID_VAL w/ assignment ) */ + this->head[stat_SY].result.sl_int += rSv(stat_GU) + rSv(stat_GN); + this->head[stat_SUM_SYS].result.sl_int += rSv(stat_GU) + rSv(stat_GN); + /* display some kinda' cpu state percentages (who or what is explained by the passed prefix) */ if (Curwin->rc.graph_cpus) {