Re: amd64/152405: wrong user time is reported when CPU frequency throttled.
The following reply was made to PR amd64/152405; it has been noted by GNATS. From: Andriy Gapon To: =?ISO-2022-JP?B?IktPSUUgSGlkZXRha2EgKBskQjhxOT4xUU40IXc/dE19NTs4JhsoQg==?= =?ISO-2022-JP?B?KSI=?= Cc: bug-follo...@freebsd.org Subject: Re: amd64/152405: wrong user time is reported when CPU frequency throttled. Date: Wed, 24 Nov 2010 08:56:26 +0200 on 24/11/2010 08:54 KOIE Hidetaka ($B8q9>1QN4!w?tM}5;8&(B) said the following: > Message-Id: <4cebe965.5050...@freebsd.org> > Date: Tue, 23 Nov 2010 18:18:45 +0200 > From: Andriy Gapon > Subject:Re: amd64/152405: wrong user time is reported when CPU f.. > > | on 23/11/2010 17:15 KOIE Hidetaka ($B8q9>1QN4!w?tM}5;8&(B) said the > following: > | > I send sysctl kern.timecounter and /var/run/dmesg.boot. > | > | Could you please additionally provide output of sysctl dev.cpu and > sysctl machdep? > > output of sysctl dev.cpu machdep is attached. > > | Meanwhile, please also try your test-case after having set the following > sysctl > | kern.eventtimer.choice="i8254". > > sysctl kern.eventtimer.choice="i8254" fails. Many apologies, this should have been kern.eventtimer.timer="i8254". Could you please try that? -- Andriy Gapon ___ freebsd-amd64@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-amd64 To unsubscribe, send any mail to "freebsd-amd64-unsubscr...@freebsd.org"
Re: amd64/152405: wrong user time is reported when CPU frequency throttled.
The following reply was made to PR amd64/152405; it has been noted by GNATS. From: KOIE Hidetaka (=?iso-2022-jp?B?GyRCOHE5PjFRTjQhdz90TX01OzgmGyhC?=) To: a...@freebsd.org Cc: bug-follo...@freebsd.org Subject: Re: amd64/152405: wrong user time is reported when CPU frequency throttled. Date: Wed, 24 Nov 2010 16:17:16 +0900 (JST) Message-Id: <4cecb71a.7040...@freebsd.org> Date: Wed, 24 Nov 2010 08:56:26 +0200 From: Andriy Gapon Subject:Re: amd64/152405: wrong user time is reported when CPU f.. | on 24/11/2010 08:54 KOIE Hidetaka ($B8q9>1QN4!w?tM}5;8&(B) said the following: | > Message-Id: <4cebe965.5050...@freebsd.org> | > Date: Tue, 23 Nov 2010 18:18:45 +0200 | > From: Andriy Gapon | > Subject:Re: amd64/152405: wrong user time is reported when CPU f.. | > | > | on 23/11/2010 17:15 KOIE Hidetaka ($B8q9>1QN4!w?tM}5;8&(B) said the following: | > | > I send sysctl kern.timecounter and /var/run/dmesg.boot. | > | | > | Could you please additionally provide output of sysctl dev.cpu and sysctl machdep? | > | > output of sysctl dev.cpu machdep is attached. | > | > | Meanwhile, please also try your test-case after having set the following sysctl | > | kern.eventtimer.choice="i8254". | > | > sysctl kern.eventtimer.choice="i8254" fails. | | Many apologies, this should have been kern.eventtimer.timer="i8254". | Could you please try that? roger. % sudo sysctl kern.eventtimer.timer="i8254" Password: kern.eventtimer.timer: LAPIC -> i8254 % sudo sysctl dev.cpu.0.freq=1800 dev.cpu.0.freq: 1800 -> 1800 % time ./loop 0.000u 4.773s 0:00.08 5962.5% 0+0k 0+0io 0pf+0w % sudo sysctl dev.cpu.0.freq=1000 dev.cpu.0.freq: 1800 -> 1000 % time ./loop 0.000u 4.773s 0:00.21 2271.4% 0+0k 0+0io 0pf+0w % sleep 1 ("sleep" had returned after sysctl kern.eventtimer.timer="LAPIC" on another terminal) % NOTE: user time is 0, system time is too long, and elapsed time is too short. And, wall-clock has been slow... % ssh my-server date ; env LANG=C date Wed Nov 24 16:03:11 JST 2010 Wed Nov 24 16:01:15 JST 2010 -- KOIE Hidetaka ___ freebsd-amd64@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-amd64 To unsubscribe, send any mail to "freebsd-amd64-unsubscr...@freebsd.org"
Re: amd64/152405: wrong user time is reported when CPU frequency throttled.
The following reply was made to PR amd64/152405; it has been noted by GNATS. From: Andriy Gapon To: =?ISO-2022-JP?B?IktPSUUgSGlkZXRha2EgKBskQjhxOT4xUU40IXc/dE19NTs4JhsoQg==?= =?ISO-2022-JP?B?KSI=?= Cc: bug-follo...@freebsd.org Subject: Re: amd64/152405: wrong user time is reported when CPU frequency throttled. Date: Wed, 24 Nov 2010 11:31:38 +0200 on 24/11/2010 09:17 KOIE Hidetaka ($B8q9>1QN4!w?tM}5;8&(B) said the following: > % sudo sysctl kern.eventtimer.timer="i8254" > Password: > kern.eventtimer.timer: LAPIC -> i8254 > % sudo sysctl dev.cpu.0.freq=1800 > dev.cpu.0.freq: 1800 -> 1800 > % time ./loop > > 0.000u 4.773s 0:00.08 5962.5% 0+0k 0+0io 0pf+0w > % sudo sysctl dev.cpu.0.freq=1000 > dev.cpu.0.freq: 1800 -> 1000 > % time ./loop > > 0.000u 4.773s 0:00.21 2271.4% 0+0k 0+0io 0pf+0w > % sleep 1 > ("sleep" had returned after sysctl kern.eventtimer.timer="LAPIC" on > another terminal) > % > > NOTE: user time is 0, system time is too long, and elapsed time is too short. > > And, wall-clock has been slow... > % ssh my-server date ; env LANG=C date > Wed Nov 24 16:03:11 JST 2010 > Wed Nov 24 16:01:15 JST 2010 Can you try to boot with kern.eventtimer.timer="i8254" in loader.conf? -- Andriy Gapon ___ freebsd-amd64@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-amd64 To unsubscribe, send any mail to "freebsd-amd64-unsubscr...@freebsd.org"
Re: amd64/152405: wrong user time is reported when CPU frequency throttled.
The following reply was made to PR amd64/152405; it has been noted by GNATS. From: KOIE Hidetaka (=?iso-2022-jp?B?GyRCOHE5PjFRTjQhdz90TX01OzgmGyhC?=) To: a...@freebsd.org Cc: bug-follo...@freebsd.org Subject: Re: amd64/152405: wrong user time is reported when CPU frequency throttled. Date: Wed, 24 Nov 2010 15:54:16 +0900 (JST) Next_Part(Wed_Nov_24_15_54_16_2010_581)-- Content-Type: Text/Plain; charset=iso-2022-jp Content-Transfer-Encoding: base64 ICBNZXNzYWdlLUlkOiA8NENFQkU5NjUuNTA1MDMwNkBmcmVlYnNkLm9yZz4NCiAgRGF0ZTogICAg ICAgVHVlLCAyMyBOb3YgMjAxMCAxODoxODo0NSArMDIwMA0KICBGcm9tOiAgICAgICBBbmRyaXkg R2Fwb24gPGF2Z0BmcmVlYnNkLm9yZz4NCiAgU3ViamVjdDogICAgUmU6IGFtZDY0LzE1MjQwNTog d3JvbmcgdXNlciB0aW1lIGlzIHJlcG9ydGVkIHdoZW4gQ1BVIGYuLg0KDQogIHwgb24gMjMvMTEv MjAxMCAxNzoxNSBLT0lFIEhpZGV0YWthICgbJEI4cTk+MVFONCF3P3RNfTU7OCYbKEIpIHNhaWQg dGhlIGZvbGxvd2luZzoNCiAgfCA+IEkgc2VuZCBzeXNjdGwga2Vybi50aW1lY291bnRlciBhbmQg L3Zhci9ydW4vZG1lc2cuYm9vdC4NCiAgfCANCiAgfCBDb3VsZCB5b3UgcGxlYXNlIGFkZGl0aW9u YWxseSBwcm92aWRlIG91dHB1dCBvZiBzeXNjdGwgZGV2LmNwdSBhbmQgc3lzY3RsIG1hY2hkZXA/ DQoNCm91dHB1dCBvZiBzeXNjdGwgZGV2LmNwdSBtYWNoZGVwIGlzIGF0dGFjaGVkLg0KDQogIHwg TWVhbndoaWxlLCBwbGVhc2UgYWxzbyB0cnkgeW91ciB0ZXN0LWNhc2UgYWZ0ZXIgaGF2aW5nIHNl dCB0aGUgZm9sbG93aW5nIHN5c2N0bA0KICB8IGtlcm4uZXZlbnR0aW1lci5jaG9pY2U9Imk4MjU0 Ii4NCg0Kc3lzY3RsIGtlcm4uZXZlbnR0aW1lci5jaG9pY2U9Imk4MjU0IiBmYWlscy4NCg0KICAg ICUgc3VkbyBzeXNjdGwga2Vybi5ldmVudHRpbWVyLmNob2ljZT0iaTgyNTQiDQogICAgUGFzc3dv cmQ6DQogICAgc3lzY3RsOiBvaWQgJ2tlcm4uZXZlbnR0aW1lci5jaG9pY2UnIGlzIHJlYWQgb25s eQ0KICAgICUNCg0KImtlcm4uZXZlbnR0aW1lci5jaG9pY2U9aTgyNTQiIGluIC9ib290L2xvYWRl ci5jb25mIGhhcyBubyBlZmZlY3QuDQpJdCBzZWVtcyB0aGF0IHRoZSB2YXJpYWJsZSBpcyByZWFk LW9ubHk6DQoNCiAgICAlIGdyZXAgLXIgJ2tlcm4uZXZlbnR0aW1lci5jaG9pY2UnIC91c3Ivc3Jj LyoNCiAgICAvdXNyL3NyYy9zaGFyZS9tYW4vbWFuNy9ldmVudHRpbWVycy43Omtlcm4uZXZlbnR0 aW1lci5jaG9pY2U6IEhQRVQoNTUwKSBMQVBJQyg0MDApIGk4MjU0KDEwMCkgUlRDKDApDQogICAg L3Vzci9zcmMvc3lzL2tlcm4va2Vybl9ldC5jOnN5c2N0bF9rZXJuX2V2ZW50dGltZXJfY2hvaWNl KFNZU0NUTF9IQU5ETEVSX0FSR1MpDQogICAgL3Vzci9zcmMvc3lzL2tlcm4va2Vybl9ldC5jOiAg ICAwLCAwLCBzeXNjdGxfa2Vybl9ldmVudHRpbWVyX2Nob2ljZSwgIkEiLCAiUHJlc2VudCBldmVu dCB0aW1lcnMiKTsNCiAgICANCjxrZXJuL2tlcm5fZXQuYz4NCiAgICBTWVNDVExfUFJPQyhfa2Vy bl9ldmVudHRpbWVyLCBPSURfQVVUTywgY2hvaWNlLA0KICAgICAgICBDVExUWVBFX1NUUklORyB8 IENUTEZMQUdfUkQgfCBDVExGTEFHX01QU0FGRSwNCiAgICAgICAgMCwgMCwgc3lzY3RsX2tlcm5f ZXZlbnR0aW1lcl9jaG9pY2UsICJBIiwgIlByZXNlbnQgZXZlbnQgdGltZXJzIik7DQoNCi0tDQpL T0lFIEhpZGV0YWthDQo= Next_Part(Wed_Nov_24_15_54_16_2010_581)-- Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: base64 Content-Disposition: inline; filename="sysctl.txt" ZGV2LmNwdS4wLiVkZXNjOiBBQ1BJIENQVQ0KZGV2LmNwdS4wLiVkcml2ZXI6IGNwdQ0KZGV2LmNw dS4wLiVsb2NhdGlvbjogaGFuZGxlPVxfUFJfLlAwMDENCmRldi5jcHUuMC4lcG5waW5mbzogX0hJ RD1ub25lIF9VSUQ9MA0KZGV2LmNwdS4wLiVwYXJlbnQ6IGFjcGkwDQpkZXYuY3B1LjAudGVtcGVy YXR1cmU6IDU2LjBDDQpkZXYuY3B1LjAuZnJlcTogMTAwMA0KZGV2LmNwdS4wLmZyZXFfbGV2ZWxz OiAxODAwLzk1MDAwIDEwMDAvMzc3ODcNCmRldi5jcHUuMC5jeF9zdXBwb3J0ZWQ6IEMxLzANCmRl di5jcHUuMC5jeF9sb3dlc3Q6IEMxDQpkZXYuY3B1LjAuY3hfdXNhZ2U6IDEwMC4wMCUgbGFzdCAx OTh1cw0KZGV2LmNwdS4xLiVkZXNjOiBBQ1BJIENQVQ0KZGV2LmNwdS4xLiVkcml2ZXI6IGNwdQ0K ZGV2LmNwdS4xLiVsb2NhdGlvbjogaGFuZGxlPVxfUFJfLlAwMDINCmRldi5jcHUuMS4lcG5waW5m bzogX0hJRD1ub25lIF9VSUQ9MA0KZGV2LmNwdS4xLiVwYXJlbnQ6IGFjcGkwDQpkZXYuY3B1LjEu dGVtcGVyYXR1cmU6IDYwLjBDDQpkZXYuY3B1LjEuY3hfc3VwcG9ydGVkOiBDMS8wDQpkZXYuY3B1 LjEuY3hfbG93ZXN0OiBDMQ0KZGV2LmNwdS4xLmN4X3VzYWdlOiAxMDAuMDAlIGxhc3QgMTIxMnVz DQpkZXYuY3B1LjIuJWRlc2M6IEFDUEkgQ1BVDQpkZXYuY3B1LjIuJWRyaXZlcjogY3B1DQpkZXYu Y3B1LjIuJWxvY2F0aW9uOiBoYW5kbGU9XF9QUl8uUDAwMw0KZGV2LmNwdS4yLiVwbnBpbmZvOiBf SElEPW5vbmUgX1VJRD0wDQpkZXYuY3B1LjIuJXBhcmVudDogYWNwaTANCmRldi5jcHUuMi50ZW1w ZXJhdHVyZTogNTcuMEMNCmRldi5jcHUuMi5jeF9zdXBwb3J0ZWQ6IEMxLzANCmRldi5jcHUuMi5j eF9sb3dlc3Q6IEMxDQpkZXYuY3B1LjIuY3hfdXNhZ2U6IDEwMC4wMCUgbGFzdCA5NTh1cw0KZGV2 LmNwdS4zLiVkZXNjOiBBQ1BJIENQVQ0KZGV2LmNwdS4zLiVkcml2ZXI6IGNwdQ0KZGV2LmNwdS4z LiVsb2NhdGlvbjogaGFuZGxlPVxfUFJfLlAwMDQNCmRldi5jcHUuMy4lcG5waW5mbzogX0hJRD1u b25lIF9VSUQ9MA0KZGV2LmNwdS4zLiVwYXJlbnQ6IGFjcGkwDQpkZXYuY3B1LjMudGVtcGVyYXR1 cmU6IDU4LjBDDQpkZXYuY3B1LjMuY3hfc3VwcG9ydGVkOiBDMS8wDQpkZXYuY3B1LjMuY3hfbG93 ZXN0OiBDMQ0KZGV2LmNwdS4zLmN4X3VzYWdlOiAxMDAuMDAlIGxhc3QgMTU3NXVzDQptYWNoZGVw LmFjcGlfdGltZXJfZnJlcTogMzU3OTU0NQ0KbWFjaGRlcC5lbmFibGVfcGFuaWNfa2V5OiAwDQpt YWNoZGVwLnJ0Y19zYXZlX3BlcmlvZDogMTgwMA0KbWFjaGRlcC5hZGprZXJudHo6IC0zMjQwMA0K bWFjaGRlcC53YWxsX2Ntb3NfY2xvY2s6IDENCm1hY2hkZXAuZGlzYWJsZV9ydGNfc2V0OiAwDQpt YWNoZGVwLmRpc2FibGVfbXRycnM6IDANCm1hY2hkZXAuaWRsZTogYWNwaQ0KbWFjaGRlcC5pZGxl X2F2YWlsYWJsZTogc3BpbiwgaGx0LCBhY3BpDQptYWNoZGVwLmlkbGVfbXdhaXQ6IDENCm1hY2hk ZXAuaGx0X2NwdXM6IDANCm1hY2hkZXAucHJvdF9mYXVsdF90cmFuc2xhdGlvbjogMA0KbWFjaGRl cC5wYW5pY19vbl9ubWk6IDENCm1hY2hkZXAua2RiX29uX25taTogMQ0KbWFjaGRlcC50c2NfZnJl cTogMTAwMDAwMDAwMA0KbWFjaGRlcC5hY3BpX3Jvb3Q6IDEwMTk2NjQNCm1hY2hkZXAuaTgyNTRf ZnJlcTogMTE5MzE4Mg0K Next_Part(W
Re: amd64/152405: wrong user time is reported when CPU frequency throttled.
The following reply was made to PR amd64/152405; it has been noted by GNATS. From: Andriy Gapon To: =?UTF-8?B?IktPSUUgSGlkZXRha2EgKOmvieaxn+iLsemahu+8oOaVsOeQhuaKgOeglCki?= Cc: bug-follo...@freebsd.org Subject: Re: amd64/152405: wrong user time is reported when CPU frequency throttled. Date: Wed, 24 Nov 2010 15:54:19 +0200 on 24/11/2010 14:33 KOIE Hidetaka (é¯æ±è±éï¼ æ°çæç ) said the following: > | Can you try to boot with kern.eventtimer.timer="i8254" in loader.conf? > > Boot process hangs up. (I tried 2 times) Oh, hm, very strange. Not sure why that happens. Anyway, it seems that I missed the real cause of this problem and the excellent explanation by Bruce. Your CPU doesn't have invariant TSC and (x86) FreeBSD doesn't handle that correctly for some things like e.g. CPU time accounting that you are reporting about. This is not a problem for modern hardware which almost universally supports invariant TSC, but older models are affected. -- Andriy Gapon ___ freebsd-amd64@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-amd64 To unsubscribe, send any mail to "freebsd-amd64-unsubscr...@freebsd.org"
Re: amd64/152405: wrong user time is reported when CPU frequency throttled.
The following reply was made to PR amd64/152405; it has been noted by GNATS. From: Andriy Gapon To: =?UTF-8?B?IktPSUUgSGlkZXRha2EgKOmvieaxn+iLsemahu+8oOaVsOeQhuaKgOeglCki?= Cc: bug-follo...@freebsd.org Subject: Re: amd64/152405: wrong user time is reported when CPU frequency throttled. Date: Wed, 24 Nov 2010 16:19:24 +0200 on 24/11/2010 15:54 Andriy Gapon said the following: > Anyway, it seems that I missed the real cause of this problem and the > excellent > explanation by Bruce. Could you please try the following (work-in-progress) patch? http://people.freebsd.org/~avg/tsc-cpu_ticks.diff I expect that it should improve accuracy for your testcase, but may penalize system performance to certain degree. As mav@ has said, performance, accuracy, powerstate management - pick any two of the three. OTOH, I am not sure by how much the performance could be affected. -- Andriy Gapon ___ freebsd-amd64@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-amd64 To unsubscribe, send any mail to "freebsd-amd64-unsubscr...@freebsd.org"
[head tinderbox] failure on amd64/amd64
TB --- 2010-11-24 22:00:01 - tinderbox 2.6 running on freebsd-current.sentex.ca TB --- 2010-11-24 22:00:01 - starting HEAD tinderbox run for amd64/amd64 TB --- 2010-11-24 22:00:01 - cleaning the object tree TB --- 2010-11-24 22:00:38 - cvsupping the source tree TB --- 2010-11-24 22:00:38 - /usr/bin/csup -z -r 3 -g -L 1 -h cvsup.sentex.ca /tinderbox/HEAD/amd64/amd64/supfile TB --- 2010-11-24 22:00:53 - building world TB --- 2010-11-24 22:00:53 - MAKEOBJDIRPREFIX=/obj TB --- 2010-11-24 22:00:53 - PATH=/usr/bin:/usr/sbin:/bin:/sbin TB --- 2010-11-24 22:00:53 - TARGET=amd64 TB --- 2010-11-24 22:00:53 - TARGET_ARCH=amd64 TB --- 2010-11-24 22:00:53 - TZ=UTC TB --- 2010-11-24 22:00:53 - __MAKE_CONF=/dev/null TB --- 2010-11-24 22:00:53 - cd /src TB --- 2010-11-24 22:00:53 - /usr/bin/make -B buildworld >>> World build started on Wed Nov 24 22:00:57 UTC 2010 >>> Rebuilding the temporary build tree >>> stage 1.1: legacy release compatibility shims >>> stage 1.2: bootstrap tools >>> stage 2.1: cleaning up the object tree >>> stage 2.2: rebuilding the object tree >>> stage 2.3: build tools >>> stage 3: cross tools >>> stage 4.1: building includes >>> stage 4.2: building libraries >>> stage 4.3: make dependencies >>> stage 4.4: building everything >>> stage 5.1: building 32 bit shim libraries >>> World build completed on Thu Nov 25 00:14:19 UTC 2010 TB --- 2010-11-25 00:14:19 - generating LINT kernel config TB --- 2010-11-25 00:14:19 - cd /src/sys/amd64/conf TB --- 2010-11-25 00:14:19 - /usr/bin/make -B LINT TB --- 2010-11-25 00:14:19 - building LINT kernel TB --- 2010-11-25 00:14:19 - MAKEOBJDIRPREFIX=/obj TB --- 2010-11-25 00:14:19 - PATH=/usr/bin:/usr/sbin:/bin:/sbin TB --- 2010-11-25 00:14:19 - TARGET=amd64 TB --- 2010-11-25 00:14:19 - TARGET_ARCH=amd64 TB --- 2010-11-25 00:14:19 - TZ=UTC TB --- 2010-11-25 00:14:19 - __MAKE_CONF=/dev/null TB --- 2010-11-25 00:14:19 - cd /src TB --- 2010-11-25 00:14:19 - /usr/bin/make -B buildkernel KERNCONF=LINT >>> Kernel build for LINT started on Thu Nov 25 00:14:19 UTC 2010 >>> stage 1: configuring the kernel >>> stage 2.1: cleaning up the object tree >>> stage 2.2: rebuilding the object tree >>> stage 2.3: build tools >>> stage 3.1: making dependencies >>> stage 3.2: building everything [...] cc -c -O2 -frename-registers -pipe -fno-strict-aliasing -std=c99 -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -Wundef -Wno-pointer-sign -fformat-extensions -nostdinc -I. -I/src/sys -I/src/sys/contrib/altq -D_KERNEL -DHAVE_KERNEL_OPTION_HEADERS -include opt_global.h -fno-common -finline-limit=8000 --param inline-unit-growth=100 --param large-function-growth=1000 -DGPROF -falign-functions=16 -DGPROF4 -DGUPROF -fno-builtin -fno-omit-frame-pointer -mcmodel=kernel -mno-red-zone -mfpmath=387 -mno-sse -mno-sse2 -mno-sse3 -mno-mmx -mno-3dnow -msoft-float -fno-asynchronous-unwind-tables -ffreestanding -fstack-protector -Werror -pg -mprofiler-epilogue /src/sys/dev/usb/usb_lookup.c cc -c -O2 -frename-registers -pipe -fno-strict-aliasing -std=c99 -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -Wundef -Wno-pointer-sign -fformat-extensions -nostdinc -I. -I/src/sys -I/src/sys/contrib/altq -D_KERNEL -DHAVE_KERNEL_OPTION_HEADERS -include opt_global.h -fno-common -finline-limit=8000 --param inline-unit-growth=100 --param large-function-growth=1000 -DGPROF -falign-functions=16 -DGPROF4 -DGUPROF -fno-builtin -fno-omit-frame-pointer -mcmodel=kernel -mno-red-zone -mfpmath=387 -mno-sse -mno-sse2 -mno-sse3 -mno-mmx -mno-3dnow -msoft-float -fno-asynchronous-unwind-tables -ffreestanding -fstack-protector -Werror -pg -mprofiler-epilogue /src/sys/dev/usb/usb_mbuf.c cc -c -O2 -frename-registers -pipe -fno-strict-aliasing -std=c99 -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -Wundef -Wno-pointer-sign -fformat-extensions -nostdinc -I. -I/src/sys -I/src/sys/contrib/altq -D_KERNEL -DHAVE_KERNEL_OPTION_HEADERS -include opt_global.h -fno-common -finline-limit=8000 --param inline-unit-growth=100 --param large-function-growth=1000 -DGPROF -falign-functions=16 -DGPROF4 -DGUPROF -fno-builtin -fno-omit-frame-pointer -mcmodel=kernel -mno-red-zone -mfpmath=387 -mno-sse -mno-sse2 -mno-sse3 -mno-mmx -mno-3dnow -msoft-float -fno-asynchronous-unwind-tables -ffreestanding -fstack-protector -Werror -pg -mprofiler-epilogue /src/sys/dev/usb/usb_msctest.c cc -c -O2 -frename-registers -pipe -fno-strict-aliasing -std=c99 -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -Wundef -Wno-pointer-sign -fformat-extensions -nostdinc -I. -I/src/sys -I/src/sys/contrib/altq -D_KERNEL -DHAVE_KERNEL_OPTION_HEADERS -include opt_global.h -fno-common -finline-limit=8000 --param inline-unit-growth=100 --param larg
Re: amd64/152405: wrong user time is reported when CPU frequency throttled.
The following reply was made to PR amd64/152405; it has been noted by GNATS. From: KOIE Hidetaka (=?iso-2022-jp?B?GyRCOHE5PjFRTjQhdz90TX01OzgmGyhC?=) To: a...@freebsd.org Cc: bug-follo...@freebsd.org Subject: Re: amd64/152405: wrong user time is reported when CPU frequency throttled. Date: Thu, 25 Nov 2010 12:56:40 +0900 (JST) Message-Id: <4ced1eec.5010...@freebsd.org> Date: Wed, 24 Nov 2010 16:19:24 +0200 From: Andriy Gapon Subject:Re: amd64/152405: wrong user time is reported when CPU f.. | on 24/11/2010 15:54 Andriy Gapon said the following: | > Anyway, it seems that I missed the real cause of this problem and the excellent | > explanation by Bruce. | | Could you please try the following (work-in-progress) patch? | http://people.freebsd.org/~avg/tsc-cpu_ticks.diff on the patched kernel, time command reports correct time! % sysctl kern.eventtimer.timer kern.eventtimer.timer: LAPIC % sudo sysctl dev.cpu.0.freq=1800 Password: dev.cpu.0.freq: 1800 -> 1800 % time ./loop 4.835u 0.007s 0:04.99 96.7% 5+2785k 0+0io 0pf+0w % sudo sysctl dev.cpu.0.freq=1000 dev.cpu.0.freq: 1800 -> 1000 % time ./loop 8.754u 0.000s 0:09.12 95.9% 5+2777k 0+0io 0pf+0w % Thank you. | | I expect that it should improve accuracy for your testcase, but may penalize | system performance to certain degree. | | As mav@ has said, performance, accuracy, powerstate management - pick any two of | the three. OTOH, I am not sure by how much the performance could be affected. -- KOIE Hidetaka ___ freebsd-amd64@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-amd64 To unsubscribe, send any mail to "freebsd-amd64-unsubscr...@freebsd.org"