Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
crickets. How do we make progress in this area? (a) can we assume Andi's json format is acceptable? We would like to know this so we don't have to reformat our data more than once. (b) Would an acceptable interim resolution the 'download area' problem be to take Andi's perf: Add support for full Intel event lists v8 and change the 'download' to refer to tools/perf/event-tables/? (c) If not, given we don't know how to get us out of the current status quo, can this patchseries still be applied, given the original complaint was the size of our events-list.h (whereas power7-events-list.h is almost twice the size)? If not, patch 3/3 in this series is still valid, no matter what, and it should still be applied (let us know if we need to resubmit). Thanks, Kim On Mon, 16 Feb 2015 10:10:45 -0600 Tom Huynh tommy.xhu...@gmail.com wrote: On Mon, Feb 09, 2015 at 09:40:19PM +0100, Andi Kleen wrote: I'll NAK any external 'download area' (and I told that Andi before): tools/perf/event-tables/ or so is a good enough 'download area' with fast enough update cycles. The proposal was to put it on kernel.org, similar to how external firmware blobs are distributed. CPU event lists are data sheets, so are like firmware. They do not follow the normal kernel code licenses. They are not source code. They cannot be reviewed in the normal way. Could you provide more details about the license and review concern? How are the event list files different from hardware- specific information (e.g. reg mapping) in header files? If any 'update' of event descriptions is needed it can happen through the distro package mechanism, or via a simple 'git pull' if it's compiled directly. Lets not overengineer this with any dependence on an external site and with a separate update mechanism - lets just get the tables into tools/ and see it from there... That experiment has been already done for oprofile, didn't work very well. Please excuse my ignorance, could you say exactly what didn't work well for oprofile? Ingo's suggestion seems good to me because these event files will be transparent to the users, and it's just more convenient not having to go to a website to look for the event file that matches the machine to download. The distro package or the perf make mechanism can put these files into the appropriate directory. The users who are not perf developers won't need to know about these files. - Tom ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
Thanks for supporting the JSON format too. (c) If not, given we don't know how to get us out of the current status quo, can this patchseries still be applied, given the original complaint was the size of our events-list.h (whereas The Intel core event lists are far larger even (and will grow even more when uncore gets added) power7-events-list.h is almost twice the size)? If not, patch 3/3 in this series is still valid, no matter what, and it should still be applied (let us know if we need to resubmit). Could also just leave out the downloader for now, so that you have to get your own event file and set it up with export EVENTMAP=... That's basically the patchkit, minus one patch. -Andi ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
Well I'm tired of discussing this. I don't think what you proposed makes sense, putting 3.4MB[1] of changing blob into perf. I'll resubmit the JSON parser without the downloader. Then users have the option to get their own events and use that. If you don't like that, standard perf just has to stay with limited events and r as before, with users having to use external tools or libraries for names for more events[2][3]. -Andi [1] Current size of https://download.01.org/perfmon/ [2] ocperf in https://github.com/andikleen/pmu-tools [3] http://perfmon2.sourceforge.net/ -- a...@linux.intel.com -- Speaking for myself only. ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
* Scott Wood scottw...@freescale.com wrote: On Mon, 2015-02-09 at 21:40 +0100, Andi Kleen wrote: I'll NAK any external 'download area' (and I told that Andi before): tools/perf/event-tables/ or so is a good enough 'download area' with fast enough update cycles. The proposal was to put it on kernel.org, similar to how external firmware blobs are distributed. [...] Fortunately perf is not an external firmware blob ... [...] CPU event lists are data sheets, so are like firmware. [...] What an absolute, idiotic, nonsense argument! CPU event lists are human readable descriptions for events. If they aren't then they have no place in tooling. Treating them like firmware is as backwards as it gets. [...] They do not follow the normal kernel code licenses. They are not source code. They cannot be reviewed in the normal way. How is it different from describing registers and bits in driver header files? What does it mean to talk about a license on information, rather than the expression of information? Andi is making idiotic arguments, instead of implementing the technically sane solution. Thanks, Ingo ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
On Mon, Feb 09, 2015 at 09:40:19PM +0100, Andi Kleen wrote: I'll NAK any external 'download area' (and I told that Andi before): tools/perf/event-tables/ or so is a good enough 'download area' with fast enough update cycles. The proposal was to put it on kernel.org, similar to how external firmware blobs are distributed. CPU event lists are data sheets, so are like firmware. They do not follow the normal kernel code licenses. They are not source code. They cannot be reviewed in the normal way. Could you provide more details about the license and review concern? How are the event list files different from hardware- specific information (e.g. reg mapping) in header files? If any 'update' of event descriptions is needed it can happen through the distro package mechanism, or via a simple 'git pull' if it's compiled directly. Lets not overengineer this with any dependence on an external site and with a separate update mechanism - lets just get the tables into tools/ and see it from there... That experiment has been already done for oprofile, didn't work very well. Please excuse my ignorance, could you say exactly what didn't work well for oprofile? Ingo's suggestion seems good to me because these event files will be transparent to the users, and it's just more convenient not having to go to a website to look for the event file that matches the machine to download. The distro package or the perf make mechanism can put these files into the appropriate directory. The users who are not perf developers won't need to know about these files. - Tom ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
On Mon, 2015-02-09 at 21:40 +0100, Andi Kleen wrote: I'll NAK any external 'download area' (and I told that Andi before): tools/perf/event-tables/ or so is a good enough 'download area' with fast enough update cycles. The proposal was to put it on kernel.org, similar to how external firmware blobs are distributed. CPU event lists are data sheets, so are like firmware. They do not follow the normal kernel code licenses. They are not source code. They cannot be reviewed in the normal way. How is it different from describing registers and bits in driver header files? What does it mean to talk about a license on information, rather than the expression of information? -Scott ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
On Fri, Feb 06, 2015 at 04:43:54PM -0600, Tom Huynh wrote: arch/powerpc/perf/e6500-events-list.h | 289 ++ That's a lot of events to stuff in the kernel, would a userspace list not be more convenient? ISTR there being various discussions on providing support for that in tools/perf, Jiri? ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
I'll NAK any external 'download area' (and I told that Andi before): tools/perf/event-tables/ or so is a good enough 'download area' with fast enough update cycles. The proposal was to put it on kernel.org, similar to how external firmware blobs are distributed. CPU event lists are data sheets, so are like firmware. They do not follow the normal kernel code licenses. They are not source code. They cannot be reviewed in the normal way. If any 'update' of event descriptions is needed it can happen through the distro package mechanism, or via a simple 'git pull' if it's compiled directly. Lets not overengineer this with any dependence on an external site and with a separate update mechanism - lets just get the tables into tools/ and see it from there... That experiment has been already done for oprofile, didn't work very well. -Andi -- a...@linux.intel.com -- Speaking for myself only. ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
* Peter Zijlstra pet...@infradead.org wrote: On Fri, Feb 06, 2015 at 04:43:54PM -0600, Tom Huynh wrote: arch/powerpc/perf/e6500-events-list.h | 289 ++ That's a lot of events to stuff in the kernel, would a userspace list not be more convenient? ISTR there being various discussions on providing support for that in tools/perf, Jiri? As long as it's in a single well organized place in tools/, I'd be fine with that solution as well. What doesn't work very well is disjunct, disorganized, inconsistent event descriptions all across the tooling and platform landscape - putting static tables into sysfs is a marked improvement over that, despite its memory usage. Thanks, Ingo ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
* Jiri Olsa jo...@redhat.com wrote: On Mon, Feb 09, 2015 at 11:07:38AM +0100, Ingo Molnar wrote: * Peter Zijlstra pet...@infradead.org wrote: On Fri, Feb 06, 2015 at 04:43:54PM -0600, Tom Huynh wrote: arch/powerpc/perf/e6500-events-list.h | 289 ++ That's a lot of events to stuff in the kernel, would a userspace list not be more convenient? ISTR there being various discussions on providing support for that in tools/perf, Jiri? As long as it's in a single well organized place in tools/, I'd be fine with that solution as well. What doesn't work very well is disjunct, disorganized, inconsistent event descriptions all across the tooling and platform landscape - putting static tables into sysfs is a marked improvement over that, despite its memory usage. the last version is in here: http://marc.info/?l=linux-kernelm=140676269017820w=2 AFAIK Andi is setting up the download area as discussed in the thread and should repost at some point I'll NAK any external 'download area' (and I told that Andi before): tools/perf/event-tables/ or so is a good enough 'download area' with fast enough update cycles. If any 'update' of event descriptions is needed it can happen through the distro package mechanism, or via a simple 'git pull' if it's compiled directly. Lets not overengineer this with any dependence on an external site and with a separate update mechanism - lets just get the tables into tools/ and see it from there... Thanks, Ingo ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/3] perf/e6500: Make event translations available in sysfs
On Mon, Feb 09, 2015 at 11:07:38AM +0100, Ingo Molnar wrote: * Peter Zijlstra pet...@infradead.org wrote: On Fri, Feb 06, 2015 at 04:43:54PM -0600, Tom Huynh wrote: arch/powerpc/perf/e6500-events-list.h | 289 ++ That's a lot of events to stuff in the kernel, would a userspace list not be more convenient? ISTR there being various discussions on providing support for that in tools/perf, Jiri? As long as it's in a single well organized place in tools/, I'd be fine with that solution as well. What doesn't work very well is disjunct, disorganized, inconsistent event descriptions all across the tooling and platform landscape - putting static tables into sysfs is a marked improvement over that, despite its memory usage. the last version is in here: http://marc.info/?l=linux-kernelm=140676269017820w=2 AFAIK Andi is setting up the download area as discussed in the thread and should repost at some point jirka ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
[PATCH 1/3] perf/e6500: Make event translations available in sysfs
Make the perf events in e6500 available via sysfs. $ ls /sys/devices/cpu/events/ branch-instructions branch-misses cache-misses cpu-cycles instructions FSL_0_INST_CMPL FSL_1_INST_CMPL ... $ cat /sys/devices/cpu/events/cpu-cycles event=0x01 Similar to the following commits: - 1c53a270724d (perf/POWER7: Make generic event translations available in sysfs). - 886c3b2d677f (perf/POWER7: Make some POWER7 events available in sysfs) - cfe0d8ba14a1 (perf tools: Make Power7 events available for perf) Signed-off-by: Tom Huynh tom.hu...@freescale.com --- arch/powerpc/include/asm/perf_event_fsl_emb.h | 20 +- arch/powerpc/perf/core-fsl-emb.c | 12 ++ arch/powerpc/perf/e6500-events-list.h | 289 ++ arch/powerpc/perf/e6500-pmu.c | 50 - 4 files changed, 365 insertions(+), 6 deletions(-) create mode 100644 arch/powerpc/perf/e6500-events-list.h diff --git a/arch/powerpc/include/asm/perf_event_fsl_emb.h b/arch/powerpc/include/asm/perf_event_fsl_emb.h index a581654..d76124f 100644 --- a/arch/powerpc/include/asm/perf_event_fsl_emb.h +++ b/arch/powerpc/include/asm/perf_event_fsl_emb.h @@ -11,6 +11,7 @@ */ #include linux/types.h +#include linux/device.h #include asm/hw_irq.h #define MAX_HWEVENTS 6 @@ -39,7 +40,7 @@ struct fsl_emb_pmu { /* Returns event flags and PMLCb (FSL_EMB_EVENT_*) */ u64 (*xlate_event)(u64 event_id); - + const struct attribute_group**attr_groups; int n_generic; int *generic_events; int (*cache_events)[PERF_COUNT_HW_CACHE_MAX] @@ -48,3 +49,20 @@ struct fsl_emb_pmu { }; int register_fsl_emb_pmu(struct fsl_emb_pmu *); + +extern ssize_t fsl_emb_events_sysfs_show(struct device *dev, +struct device_attribute *attr, +char *page); + +#define EVENT_VAR(_id, _suffix) event_attr_##_id##_suffix +#define EVENT_PTR(_id, _suffix) (event_attr_##_id##_suffix.attr.attr) + +#define EVENT_ATTR(_name, _id, _suffix) \ + PMU_EVENT_ATTR(_name, EVENT_VAR(_id, _suffix), PME_##_id, \ + fsl_emb_events_sysfs_show) + +#define GENERIC_EVENT_ATTR(_name, _id) EVENT_ATTR(_name, _id, _g) +#define GENERIC_EVENT_PTR(_id) EVENT_PTR(_id, _g) + +#define FSL_EMB_EVENT_ATTR(_name, _id) EVENT_ATTR(_name, _id, _p) +#define FSL_EMB_EVENT_PTR(_id) EVENT_PTR(_id, _p) diff --git a/arch/powerpc/perf/core-fsl-emb.c b/arch/powerpc/perf/core-fsl-emb.c index 4acaea0..d4af2f3 100644 --- a/arch/powerpc/perf/core-fsl-emb.c +++ b/arch/powerpc/perf/core-fsl-emb.c @@ -581,6 +581,16 @@ static int fsl_emb_pmu_event_init(struct perf_event *event) return err; } +ssize_t fsl_emb_events_sysfs_show(struct device *dev, + struct device_attribute *attr, char *page) +{ + struct perf_pmu_events_attr *pmu_attr; + + pmu_attr = container_of(attr, struct perf_pmu_events_attr, attr); + + return sprintf(page, event=0x%02llx\n, pmu_attr-id); +} + static struct pmu fsl_emb_pmu = { .pmu_enable = fsl_emb_pmu_enable, .pmu_disable= fsl_emb_pmu_disable, @@ -711,6 +721,8 @@ int register_fsl_emb_pmu(struct fsl_emb_pmu *pmu) pr_info(%s performance monitor hardware support registered\n, pmu-name); + fsl_emb_pmu.attr_groups = ppmu-attr_groups; + perf_pmu_register(fsl_emb_pmu, cpu, PERF_TYPE_RAW); return 0; diff --git a/arch/powerpc/perf/e6500-events-list.h b/arch/powerpc/perf/e6500-events-list.h new file mode 100644 index 000..27be38a --- /dev/null +++ b/arch/powerpc/perf/e6500-events-list.h @@ -0,0 +1,289 @@ +/* + * Performance counter support for e6500 family processors. + * + * Copyright 2015 Freescale Semiconductor, Inc. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ +EVENT(FSL_CYC, 1) +EVENT(FSL_INST_CMPL, 2) +EVENT(FSL_UOPS_CMPL, 3) +EVENT(FSL_UOPS_DECODED,5) +EVENT(FSL_PM_EV_TRANS, 6) +EVENT(FSL_PM_EV_CYC, 7) +EVENT(FSL_BR_CMPL, 8) +EVENT(FSL_BR_TKN, 67) +EVENT(FSL_BLR_TKN, 68) +EVENT(FSL_NUM_CQ_REDIR,11) +EVENT(FSL_LD_UOPS_CMPL,9) +EVENT(FSL_ST_UOPS_CMPL,10) +EVENT(FSL_LSU_UOPS_CMPL, 181) +EVENT(FSL_GPR_LD_CMPL,