Re: [Intel-gfx] [PATCH i-g-t 01/11] lib: Helper library for parsing i915 fdinfo output

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




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@xxxxxxxxx>

Tests and intel_gpu_top will share common code for parsing this file.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
---
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




[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux