From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ingo Molnar Subject: Re: [GIT PULL] perf/urgent fixes Date: Thu, 7 Nov 2019 08:02:29 +0100 Message-ID: <20191107070229.GA30739@gmail.com> References: <20191017160301.20888-1-acme@kernel.org> <20191021062354.GA22042@gmail.com> <20191106191051.GE3636@kernel.org> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit Return-path: Content-Disposition: inline In-Reply-To: <20191106191051.GE3636@kernel.org> Sender: linux-kernel-owner@vger.kernel.org To: Arnaldo Carvalho de Melo Cc: Thomas Gleixner , Jiri Olsa , Namhyung Kim , Clark Williams , linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org, Adrian Hunter , Andi Kleen , "Gustavo A . R . Silva" , Thomas Richter , Yunfeng Ye , Arnaldo Carvalho de Melo List-Id: linux-perf-users.vger.kernel.org * Arnaldo Carvalho de Melo wrote: > Em Mon, Oct 21, 2019 at 08:23:54AM +0200, Ingo Molnar escreveu: > > Secondly, there also appears to be a TUI weirdness when the annotated > > kernel functions are small (or weird): the blue cursor is stuck at the > > top and I cannot move between the annotated instructions with the down/up > > arrow: > > > > Samples: 13M of event 'cycles', 4000 Hz, Event count (approx.): 1272420588851 > > clear_page_rep /usr/lib/debug/boot/vmlinux-5.4.0-rc3-custom-00557-gb6c81ae120e0 [Percent: local period] > > 0.01 │ mov $0x200,%ecx ▒ > > │ xorl %eax,%eax ▒ > > 0.01 │ xor %eax,%eax ▒ > > │ rep stosq ▒ > > 99.27 │ rep stos %rax,%es:(%rdi) ▒ > > │ ret ▒ > > 0.71 │ ← retq > > > > I can still exit the screen with 'q', and can move around in larger > > annotated kernel functions. Not sure whether it's related to function > > size, or perhaps to the 'hottest' instruction that the cursor is normally > > placed at. > > I couldn't reproduce this one so far, with another small function, > clear_page_erms, what happens is that the cursor seems to be hidden at the > bottom, if you press the "magic" 'D' hotkey it will tell (at the bottom of the > screen) that the idx is at 18, which for a function with just 8 output lines > doesn't make sense: > > Samples: 12K of event 'cycles', 4000 Hz, Event count (approx.): 2219443843 > clear_page_erms /proc/kcore [Percent: local period] > Percent│ ◆ > │ ▒ > │ ▒ > │ Disassembly of section load0: ▒ > │ ▒ > │ ffffffffab9d17c0 : ▒ > │ mov $0x1000,%ecx ▒ > │ xor %eax,%eax ▒ > 100.00 │ rep stos %al,%es:(%rdi) ▒ > │ ← retq ▒ > ▒ > 1: nr_ent=20, height=35, idx=18, top_idx=1, nr_asm_entries=8 > > Doesn't make sense, there aren't 20 entries nor the idx is 18, the > nr_asm_entries is right, I'll try to follow up on this one... > > But you can in these cases try to go on pressin the up arrow till the cursor > appears, etc. Yeah, I can reproduce this too - and don't see the original weirdness anymore. :-/ Will re-report if it happens again. Thanks, Ingo