All of lore.kernel.org
 help / color / mirror / Atom feed
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Cc: igt-dev@lists.freedesktop.org, Intel-gfx@lists.freedesktop.org,
	dri-devel@lists.freedesktop.org
Subject: Re: [Intel-gfx] [PATCH i-g-t 01/11] lib: Helper library for parsing i915 fdinfo output
Date: Thu, 31 Mar 2022 14:22:34 +0100	[thread overview]
Message-ID: <580836e7-0ed9-6d1e-c3ff-a0798009293b@linux.intel.com> (raw)
In-Reply-To: <20220330195229.GD59408@unerlige-ril-10.165.21.154>


On 30/03/2022 20:52, Umesh Nerlige Ramappa wrote:
> On Tue, Feb 22, 2022 at 01:55:55PM +0000, Tvrtko Ursulin wrote:
>> From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>>
>> Tests and intel_gpu_top will share common code for parsing this file.
>>
>> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>> ---
>> lib/igt_drm_fdinfo.c | 183 +++++++++++++++++++++++++++++++++++++++++++
>> lib/igt_drm_fdinfo.h |  48 ++++++++++++
>> lib/meson.build      |   7 ++
>> 3 files changed, 238 insertions(+)
>> create mode 100644 lib/igt_drm_fdinfo.c
>> create mode 100644 lib/igt_drm_fdinfo.h
>>
>> diff --git a/lib/igt_drm_fdinfo.c b/lib/igt_drm_fdinfo.c
>> new file mode 100644
>> index 000000000000..28c1bdbda08e
>> --- /dev/null
>> +++ b/lib/igt_drm_fdinfo.c
>> @@ -0,0 +1,183 @@
>> +/*
>> + * Copyright © 2022 Intel Corporation
>> + *
>> + * Permission is hereby granted, free of charge, to any person 
>> obtaining a
>> + * copy of this software and associated documentation files (the 
>> "Software"),
>> + * to deal in the Software without restriction, including without 
>> limitation
>> + * the rights to use, copy, modify, merge, publish, distribute, 
>> sublicense,
>> + * and/or sell copies of the Software, and to permit persons to whom the
>> + * Software is furnished to do so, subject to the following conditions:
>> + *
>> + * The above copyright notice and this permission notice (including 
>> the next
>> + * paragraph) shall be included in all copies or substantial portions 
>> of the
>> + * Software.
>> + *
>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 
>> EXPRESS OR
>> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 
>> MERCHANTABILITY,
>> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT 
>> SHALL
>> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES 
>> OR OTHER
>> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 
>> ARISING
>> + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 
>> OTHER DEALINGS
>> + * IN THE SOFTWARE.
>> + *
>> + */
>> +
>> +#include <ctype.h>
>> +#include <sys/types.h>
>> +#include <sys/stat.h>
>> +#include <fcntl.h>
>> +#include <stdio.h>
>> +#include <string.h>
>> +#include <stdlib.h>
>> +#include <unistd.h>
>> +
>> +#include "drmtest.h"
>> +
>> +#include "igt_drm_fdinfo.h"
>> +
>> +static size_t read_fdinfo(char *buf, const size_t sz, int at, const 
>> char *name)
>> +{
>> +    size_t count;
>> +    int fd;
>> +
>> +    fd = openat(at, name, O_RDONLY);
>> +    if (fd < 0)
>> +        return 0;
>> +
>> +    buf[sz - 1] = 0;
> 
> Wondering if this ^ should be after the read() in case 4096 bytes are read.

Yes, I guess even before and after so it's guaranteed to be null 
terminated no matter what.

> 
>> +    count = read(fd, buf, sz);
>> +    close(fd);
>> +
>> +    return count;
>> +}
>> +
>> +static int parse_engine(char *line, struct drm_client_fdinfo *info,
>> +            size_t prefix_len, uint64_t *val)
>> +{
>> +    static const char *e2class[] = {
>> +        "render",
>> +        "copy",
>> +        "video",
>> +        "video-enhance",
>> +    };
>> +    ssize_t name_len;
>> +    char *name, *p;
>> +    int found = -1;
>> +    unsigned int i;
>> +
>> +    p = index(line, ':');
>> +    if (!p || p == line)
>> +        return -1;
>> +
>> +    name_len = p - line - prefix_len;
>> +    if (name_len < 1)
>> +        return -1;
>> +
>> +    name = line + prefix_len;
>> +
>> +    for (i = 0; i < ARRAY_SIZE(e2class); i++) {
>> +        if (!strncmp(name, e2class[i], name_len)) {
>> +            found = i;
>> +            break;
>> +        }
>> +    }
>> +
>> +    if (found >= 0) {
>> +        while (*++p && isspace(*p));
>> +        *val = strtoull(p, NULL, 10);
>> +    }
>> +
>> +    return found;
>> +}
>> +
>> +static const char *find_kv(const char *buf, const char *key, size_t 
>> keylen)
>> +{
>> +    const char *p = buf;
>> +
>> +    p = index(buf, ':');
>> +    if (!p || p == buf)
>> +        return NULL;
>> +
>> +    if ((p - buf) != keylen)
>> +        return NULL;
>> +
>> +    while (*++p && isspace(*p));
>> +    if (*p && !strncmp(buf, key, keylen))
> 
> nit: why not just do the strncmp early in this function since buf, key, 
> keylen have not changed?

I see what you mean, first check that the key/prefix matches and do the 
other checks, right?

I honestly don't remember if there was a special reason why I laid it 
out like it is. I know it is an annoyingly hot path and I did do many 
tweaks to bring the CPU utilisation low.

But thinking about it now, the strncmp at the start should indeed be 
more efficient so I will change it.

> 
>> +        return p;
>> +
>> +    return NULL;
>> +}
>> +
>> +bool
>> +__igt_parse_drm_fdinfo(int dir, const char *fd, struct 
>> drm_client_fdinfo *info)
>> +{
>> +    char buf[4096], *_buf = buf;
>> +    char *l, *ctx = NULL;
>> +    unsigned int good = 0;
>> +    size_t count;
>> +
> 
> Should buf be zeroed out here?

Right, either here or in read_fdinfo as mentioned above.

>> +    count = read_fdinfo(buf, sizeof(buf), dir, fd);
>> +    if (!count)
>> +        return false;
>> +
>> +    while ((l = strtok_r(_buf, "\n", &ctx))) {
>> +        uint64_t val = 0;
>> +        const char *v;
>> +        int idx;
>> +
>> +        _buf = NULL;
>> +
>> +        if ((v = find_kv(l, "drm-driver", strlen("drm-driver")))) {
>> +            strncpy(info->driver, v, sizeof(info->driver) - 1);
>> +            good++;
>> +        } else if ((v = find_kv(l, "drm-pdev", strlen("drm-pdev")))) {
>> +            strncpy(info->pdev, v, sizeof(info->pdev) - 1);
>> +        }  else if ((v = find_kv(l, "drm-client-id",
>> +                     strlen("drm-client-id")))) {
>> +            info->id = atol(v);
>> +            good++;
>> +        } else if (!strncmp(l, "drm-engine-", 11) &&
>> +               strncmp(l, "drm-engine-capacity-", 20)) {
>> +            idx = parse_engine(l, info, strlen("drm-engine-"),
>> +                       &val);
>> +            if (idx >= 0) {
>> +                if (!info->capacity[idx])
>> +                    info->capacity[idx] = 1;
>> +                info->busy[idx] = val;
>> +                info->num_engines++;
>> +            }
>> +        } else if (!strncmp(l, "drm-engine-capacity-", 20)) {
>> +            idx = parse_engine(l, info,
>> +                       strlen("drm-engine-capacity-"),
>> +                       &val);
>> +            if (idx >= 0)
>> +                info->capacity[idx] = val;
>> +        }
>> +    }
>> +
>> +    if (good < 2 || !info->num_engines)
>> +        return false; /* fdinfo format not as expected */
>> +
>> +    return true;
>> +}
>> +
>> +bool igt_parse_drm_fdinfo(int drm_fd, struct drm_client_fdinfo *info)
>> +{
>> +    char fd[64];
>> +    int dir, ret;
>> +    bool res;
>> +
>> +    ret = snprintf(fd, sizeof(fd), "%u", drm_fd);
>> +    if (ret < 0 || ret == sizeof(fd))
>> +        return false;
>> +
>> +    dir = open("/proc/self/fdinfo", O_DIRECTORY | O_RDONLY);
>> +    if (dir < 0)
>> +        return false;
>> +
>> +    res = __igt_parse_drm_fdinfo(dir, fd, info);
>> +
>> +    close(dir);
>> +
>> +    return res;
>> +}
>> diff --git a/lib/igt_drm_fdinfo.h b/lib/igt_drm_fdinfo.h
>> new file mode 100644
>> index 000000000000..c527bab9a204
>> --- /dev/null
>> +++ b/lib/igt_drm_fdinfo.h
>> @@ -0,0 +1,48 @@
>> +/*
>> + * Copyright © 2022 Intel Corporation
>> + *
>> + * Permission is hereby granted, free of charge, to any person 
>> obtaining a
>> + * copy of this software and associated documentation files (the 
>> "Software"),
>> + * to deal in the Software without restriction, including without 
>> limitation
>> + * the rights to use, copy, modify, merge, publish, distribute, 
>> sublicense,
>> + * and/or sell copies of the Software, and to permit persons to whom the
>> + * Software is furnished to do so, subject to the following conditions:
>> + *
>> + * The above copyright notice and this permission notice (including 
>> the next
>> + * paragraph) shall be included in all copies or substantial portions 
>> of the
>> + * Software.
>> + *
>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 
>> EXPRESS OR
>> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 
>> MERCHANTABILITY,
>> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT 
>> SHALL
>> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES 
>> OR OTHER
>> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 
>> ARISING
>> + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 
>> OTHER DEALINGS
>> + * IN THE SOFTWARE.
>> + *
>> + */
>> +
>> +#ifndef IGT_DRM_FDINFO_H
>> +#define IGT_DRM_FDINFO_H
>> +
>> +#include <sys/types.h>
>> +#include <dirent.h>
>> +#include <stdint.h>
>> +#include <stdbool.h>
>> +
>> +struct drm_client_fdinfo {
>> +    char driver[128];
>> +    char pdev[128];
>> +    unsigned long id;
>> +
>> +    unsigned int num_engines;
>> +    unsigned int capacity[16];
>> +    uint64_t busy[16];
> 
> I see 16 in a lot of places, wondering if it could use a #define?

I can see them here and one mention in igt_drm_clients. The latter it 
looks I can un-export igt_drm_client_update and make it take struct 
drm_client_fdinfo to better hide this.

And for here I will add a define like DRM_CLIENT_FDINFO_MAX_ENGINES.

Regards,

Tvrtko

> 
> Thanks,
> Umesh
> 
>> +};
>> +
>> +bool igt_parse_drm_fdinfo(int drm_fd, struct drm_client_fdinfo *info);
>> +
>> +bool __igt_parse_drm_fdinfo(int dir, const char *fd,
>> +                struct drm_client_fdinfo *info);
>> +
>> +#endif /* IGT_DRM_FDINFO_H */
>> diff --git a/lib/meson.build b/lib/meson.build
>> index 3e43316d1e36..d00817dc58bc 100644
>> --- a/lib/meson.build
>> +++ b/lib/meson.build
>> @@ -16,6 +16,7 @@ lib_sources = [
>>     'igt_debugfs.c',
>>     'igt_device.c',
>>     'igt_device_scan.c',
>> +    'igt_drm_fdinfo.c',
>>     'igt_aux.c',
>>     'igt_gt.c',
>>     'igt_halffloat.c',
>> @@ -216,6 +217,12 @@ lib_igt_device_scan_build = 
>> static_library('igt_device_scan',
>> lib_igt_device_scan = declare_dependency(link_with : 
>> lib_igt_device_scan_build,
>>                   include_directories : inc)
>>
>> +lib_igt_drm_fdinfo_build = static_library('igt_drm_fdinfo',
>> +    ['igt_drm_fdinfo.c'],
>> +    include_directories : inc)
>> +
>> +lib_igt_drm_fdinfo = declare_dependency(link_with : 
>> lib_igt_drm_fdinfo_build,
>> +                  include_directories : inc)
>> i915_perf_files = [
>>   'igt_list.c',
>>   'i915/perf.c',
>> -- 
>> 2.32.0
>>

WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Cc: igt-dev@lists.freedesktop.org, Intel-gfx@lists.freedesktop.org,
	dri-devel@lists.freedesktop.org
Subject: Re: [igt-dev] [Intel-gfx] [PATCH i-g-t 01/11] lib: Helper library for parsing i915 fdinfo output
Date: Thu, 31 Mar 2022 14:22:34 +0100	[thread overview]
Message-ID: <580836e7-0ed9-6d1e-c3ff-a0798009293b@linux.intel.com> (raw)
In-Reply-To: <20220330195229.GD59408@unerlige-ril-10.165.21.154>


On 30/03/2022 20:52, Umesh Nerlige Ramappa wrote:
> On Tue, Feb 22, 2022 at 01:55:55PM +0000, Tvrtko Ursulin wrote:
>> From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>>
>> Tests and intel_gpu_top will share common code for parsing this file.
>>
>> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>> ---
>> lib/igt_drm_fdinfo.c | 183 +++++++++++++++++++++++++++++++++++++++++++
>> lib/igt_drm_fdinfo.h |  48 ++++++++++++
>> lib/meson.build      |   7 ++
>> 3 files changed, 238 insertions(+)
>> create mode 100644 lib/igt_drm_fdinfo.c
>> create mode 100644 lib/igt_drm_fdinfo.h
>>
>> diff --git a/lib/igt_drm_fdinfo.c b/lib/igt_drm_fdinfo.c
>> new file mode 100644
>> index 000000000000..28c1bdbda08e
>> --- /dev/null
>> +++ b/lib/igt_drm_fdinfo.c
>> @@ -0,0 +1,183 @@
>> +/*
>> + * Copyright © 2022 Intel Corporation
>> + *
>> + * Permission is hereby granted, free of charge, to any person 
>> obtaining a
>> + * copy of this software and associated documentation files (the 
>> "Software"),
>> + * to deal in the Software without restriction, including without 
>> limitation
>> + * the rights to use, copy, modify, merge, publish, distribute, 
>> sublicense,
>> + * and/or sell copies of the Software, and to permit persons to whom the
>> + * Software is furnished to do so, subject to the following conditions:
>> + *
>> + * The above copyright notice and this permission notice (including 
>> the next
>> + * paragraph) shall be included in all copies or substantial portions 
>> of the
>> + * Software.
>> + *
>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 
>> EXPRESS OR
>> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 
>> MERCHANTABILITY,
>> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT 
>> SHALL
>> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES 
>> OR OTHER
>> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 
>> ARISING
>> + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 
>> OTHER DEALINGS
>> + * IN THE SOFTWARE.
>> + *
>> + */
>> +
>> +#include <ctype.h>
>> +#include <sys/types.h>
>> +#include <sys/stat.h>
>> +#include <fcntl.h>
>> +#include <stdio.h>
>> +#include <string.h>
>> +#include <stdlib.h>
>> +#include <unistd.h>
>> +
>> +#include "drmtest.h"
>> +
>> +#include "igt_drm_fdinfo.h"
>> +
>> +static size_t read_fdinfo(char *buf, const size_t sz, int at, const 
>> char *name)
>> +{
>> +    size_t count;
>> +    int fd;
>> +
>> +    fd = openat(at, name, O_RDONLY);
>> +    if (fd < 0)
>> +        return 0;
>> +
>> +    buf[sz - 1] = 0;
> 
> Wondering if this ^ should be after the read() in case 4096 bytes are read.

Yes, I guess even before and after so it's guaranteed to be null 
terminated no matter what.

> 
>> +    count = read(fd, buf, sz);
>> +    close(fd);
>> +
>> +    return count;
>> +}
>> +
>> +static int parse_engine(char *line, struct drm_client_fdinfo *info,
>> +            size_t prefix_len, uint64_t *val)
>> +{
>> +    static const char *e2class[] = {
>> +        "render",
>> +        "copy",
>> +        "video",
>> +        "video-enhance",
>> +    };
>> +    ssize_t name_len;
>> +    char *name, *p;
>> +    int found = -1;
>> +    unsigned int i;
>> +
>> +    p = index(line, ':');
>> +    if (!p || p == line)
>> +        return -1;
>> +
>> +    name_len = p - line - prefix_len;
>> +    if (name_len < 1)
>> +        return -1;
>> +
>> +    name = line + prefix_len;
>> +
>> +    for (i = 0; i < ARRAY_SIZE(e2class); i++) {
>> +        if (!strncmp(name, e2class[i], name_len)) {
>> +            found = i;
>> +            break;
>> +        }
>> +    }
>> +
>> +    if (found >= 0) {
>> +        while (*++p && isspace(*p));
>> +        *val = strtoull(p, NULL, 10);
>> +    }
>> +
>> +    return found;
>> +}
>> +
>> +static const char *find_kv(const char *buf, const char *key, size_t 
>> keylen)
>> +{
>> +    const char *p = buf;
>> +
>> +    p = index(buf, ':');
>> +    if (!p || p == buf)
>> +        return NULL;
>> +
>> +    if ((p - buf) != keylen)
>> +        return NULL;
>> +
>> +    while (*++p && isspace(*p));
>> +    if (*p && !strncmp(buf, key, keylen))
> 
> nit: why not just do the strncmp early in this function since buf, key, 
> keylen have not changed?

I see what you mean, first check that the key/prefix matches and do the 
other checks, right?

I honestly don't remember if there was a special reason why I laid it 
out like it is. I know it is an annoyingly hot path and I did do many 
tweaks to bring the CPU utilisation low.

But thinking about it now, the strncmp at the start should indeed be 
more efficient so I will change it.

> 
>> +        return p;
>> +
>> +    return NULL;
>> +}
>> +
>> +bool
>> +__igt_parse_drm_fdinfo(int dir, const char *fd, struct 
>> drm_client_fdinfo *info)
>> +{
>> +    char buf[4096], *_buf = buf;
>> +    char *l, *ctx = NULL;
>> +    unsigned int good = 0;
>> +    size_t count;
>> +
> 
> Should buf be zeroed out here?

Right, either here or in read_fdinfo as mentioned above.

>> +    count = read_fdinfo(buf, sizeof(buf), dir, fd);
>> +    if (!count)
>> +        return false;
>> +
>> +    while ((l = strtok_r(_buf, "\n", &ctx))) {
>> +        uint64_t val = 0;
>> +        const char *v;
>> +        int idx;
>> +
>> +        _buf = NULL;
>> +
>> +        if ((v = find_kv(l, "drm-driver", strlen("drm-driver")))) {
>> +            strncpy(info->driver, v, sizeof(info->driver) - 1);
>> +            good++;
>> +        } else if ((v = find_kv(l, "drm-pdev", strlen("drm-pdev")))) {
>> +            strncpy(info->pdev, v, sizeof(info->pdev) - 1);
>> +        }  else if ((v = find_kv(l, "drm-client-id",
>> +                     strlen("drm-client-id")))) {
>> +            info->id = atol(v);
>> +            good++;
>> +        } else if (!strncmp(l, "drm-engine-", 11) &&
>> +               strncmp(l, "drm-engine-capacity-", 20)) {
>> +            idx = parse_engine(l, info, strlen("drm-engine-"),
>> +                       &val);
>> +            if (idx >= 0) {
>> +                if (!info->capacity[idx])
>> +                    info->capacity[idx] = 1;
>> +                info->busy[idx] = val;
>> +                info->num_engines++;
>> +            }
>> +        } else if (!strncmp(l, "drm-engine-capacity-", 20)) {
>> +            idx = parse_engine(l, info,
>> +                       strlen("drm-engine-capacity-"),
>> +                       &val);
>> +            if (idx >= 0)
>> +                info->capacity[idx] = val;
>> +        }
>> +    }
>> +
>> +    if (good < 2 || !info->num_engines)
>> +        return false; /* fdinfo format not as expected */
>> +
>> +    return true;
>> +}
>> +
>> +bool igt_parse_drm_fdinfo(int drm_fd, struct drm_client_fdinfo *info)
>> +{
>> +    char fd[64];
>> +    int dir, ret;
>> +    bool res;
>> +
>> +    ret = snprintf(fd, sizeof(fd), "%u", drm_fd);
>> +    if (ret < 0 || ret == sizeof(fd))
>> +        return false;
>> +
>> +    dir = open("/proc/self/fdinfo", O_DIRECTORY | O_RDONLY);
>> +    if (dir < 0)
>> +        return false;
>> +
>> +    res = __igt_parse_drm_fdinfo(dir, fd, info);
>> +
>> +    close(dir);
>> +
>> +    return res;
>> +}
>> diff --git a/lib/igt_drm_fdinfo.h b/lib/igt_drm_fdinfo.h
>> new file mode 100644
>> index 000000000000..c527bab9a204
>> --- /dev/null
>> +++ b/lib/igt_drm_fdinfo.h
>> @@ -0,0 +1,48 @@
>> +/*
>> + * Copyright © 2022 Intel Corporation
>> + *
>> + * Permission is hereby granted, free of charge, to any person 
>> obtaining a
>> + * copy of this software and associated documentation files (the 
>> "Software"),
>> + * to deal in the Software without restriction, including without 
>> limitation
>> + * the rights to use, copy, modify, merge, publish, distribute, 
>> sublicense,
>> + * and/or sell copies of the Software, and to permit persons to whom the
>> + * Software is furnished to do so, subject to the following conditions:
>> + *
>> + * The above copyright notice and this permission notice (including 
>> the next
>> + * paragraph) shall be included in all copies or substantial portions 
>> of the
>> + * Software.
>> + *
>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 
>> EXPRESS OR
>> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 
>> MERCHANTABILITY,
>> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT 
>> SHALL
>> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES 
>> OR OTHER
>> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 
>> ARISING
>> + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 
>> OTHER DEALINGS
>> + * IN THE SOFTWARE.
>> + *
>> + */
>> +
>> +#ifndef IGT_DRM_FDINFO_H
>> +#define IGT_DRM_FDINFO_H
>> +
>> +#include <sys/types.h>
>> +#include <dirent.h>
>> +#include <stdint.h>
>> +#include <stdbool.h>
>> +
>> +struct drm_client_fdinfo {
>> +    char driver[128];
>> +    char pdev[128];
>> +    unsigned long id;
>> +
>> +    unsigned int num_engines;
>> +    unsigned int capacity[16];
>> +    uint64_t busy[16];
> 
> I see 16 in a lot of places, wondering if it could use a #define?

I can see them here and one mention in igt_drm_clients. The latter it 
looks I can un-export igt_drm_client_update and make it take struct 
drm_client_fdinfo to better hide this.

And for here I will add a define like DRM_CLIENT_FDINFO_MAX_ENGINES.

Regards,

Tvrtko

> 
> Thanks,
> Umesh
> 
>> +};
>> +
>> +bool igt_parse_drm_fdinfo(int drm_fd, struct drm_client_fdinfo *info);
>> +
>> +bool __igt_parse_drm_fdinfo(int dir, const char *fd,
>> +                struct drm_client_fdinfo *info);
>> +
>> +#endif /* IGT_DRM_FDINFO_H */
>> diff --git a/lib/meson.build b/lib/meson.build
>> index 3e43316d1e36..d00817dc58bc 100644
>> --- a/lib/meson.build
>> +++ b/lib/meson.build
>> @@ -16,6 +16,7 @@ lib_sources = [
>>     'igt_debugfs.c',
>>     'igt_device.c',
>>     'igt_device_scan.c',
>> +    'igt_drm_fdinfo.c',
>>     'igt_aux.c',
>>     'igt_gt.c',
>>     'igt_halffloat.c',
>> @@ -216,6 +217,12 @@ lib_igt_device_scan_build = 
>> static_library('igt_device_scan',
>> lib_igt_device_scan = declare_dependency(link_with : 
>> lib_igt_device_scan_build,
>>                   include_directories : inc)
>>
>> +lib_igt_drm_fdinfo_build = static_library('igt_drm_fdinfo',
>> +    ['igt_drm_fdinfo.c'],
>> +    include_directories : inc)
>> +
>> +lib_igt_drm_fdinfo = declare_dependency(link_with : 
>> lib_igt_drm_fdinfo_build,
>> +                  include_directories : inc)
>> i915_perf_files = [
>>   'igt_list.c',
>>   'i915/perf.c',
>> -- 
>> 2.32.0
>>

  reply	other threads:[~2022-03-31 13:22 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-02-22 13:55 [PATCH i-g-t 00/11] Per client GPU utilisation Tvrtko Ursulin
2022-02-22 13:55 ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:55 ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:55 ` [PATCH i-g-t 01/11] lib: Helper library for parsing i915 fdinfo output Tvrtko Ursulin
2022-02-22 13:55   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:55   ` [Intel-gfx] " Tvrtko Ursulin
2022-03-30 19:52   ` Umesh Nerlige Ramappa
2022-03-30 19:52     ` [igt-dev] " Umesh Nerlige Ramappa
2022-03-31 13:22     ` Tvrtko Ursulin [this message]
2022-03-31 13:22       ` Tvrtko Ursulin
2022-02-22 13:55 ` [Intel-gfx] [PATCH i-g-t 02/11] tests/i915/drm_fdinfo: Basic and functional tests for GPU busyness exported via fdinfo Tvrtko Ursulin
2022-02-22 13:55   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:55   ` Tvrtko Ursulin
2022-03-30 20:11   ` [igt-dev] " Umesh Nerlige Ramappa
2022-03-30 20:11     ` [Intel-gfx] " Umesh Nerlige Ramappa
2022-03-31 13:23     ` Tvrtko Ursulin
2022-03-31 13:23       ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:55 ` [PATCH i-g-t 03/11] intel-gpu-top: Add support for per client stats Tvrtko Ursulin
2022-02-22 13:55   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:55   ` [Intel-gfx] " Tvrtko Ursulin
2022-03-31 22:08   ` [igt-dev] " Umesh Nerlige Ramappa
2022-03-31 22:08     ` [Intel-gfx] " Umesh Nerlige Ramappa
2022-04-01 14:19     ` Tvrtko Ursulin
2022-04-01 14:19       ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:55 ` [PATCH i-g-t 04/11] lib: Extract igt_drm_clients from intel_gpu_top Tvrtko Ursulin
2022-02-22 13:55   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:55   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:55 ` [PATCH i-g-t 05/11] libdrmfdinfo: Allow specifying custom engine map Tvrtko Ursulin
2022-02-22 13:55   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:55   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:56 ` [PATCH i-g-t 06/11] libdrmclients: Record client drm minor Tvrtko Ursulin
2022-02-22 13:56   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:56   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:56 ` [PATCH i-g-t 07/11] libdrmclient: Support multiple DRM cards Tvrtko Ursulin
2022-02-22 13:56   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:56   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:56 ` [PATCH i-g-t 08/11] libdrmfdinfo: Track largest engine index Tvrtko Ursulin
2022-02-22 13:56   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:56   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:56 ` [PATCH i-g-t 09/11] libdrmclient/intel_gpu_top: Decouple hardcoded engine assumptions Tvrtko Ursulin
2022-02-22 13:56   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:56   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:56 ` [PATCH i-g-t 10/11] libdrmclient: Enforce client status sort order in the library Tvrtko Ursulin
2022-02-22 13:56   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:56   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 13:56 ` [PATCH i-g-t 11/11] gputop: Basic vendor agnostic GPU top tool Tvrtko Ursulin
2022-02-22 13:56   ` [igt-dev] " Tvrtko Ursulin
2022-02-22 13:56   ` [Intel-gfx] " Tvrtko Ursulin
2022-02-22 17:06 ` [igt-dev] ✓ Fi.CI.BAT: success for Per client GPU utilisation Patchwork
2022-02-22 21:17 ` [igt-dev] ✓ Fi.CI.IGT: " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=580836e7-0ed9-6d1e-c3ff-a0798009293b@linux.intel.com \
    --to=tvrtko.ursulin@linux.intel.com \
    --cc=Intel-gfx@lists.freedesktop.org \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=igt-dev@lists.freedesktop.org \
    --cc=umesh.nerlige.ramappa@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.