Re: [PATCH v4 06/22] perf map: Move map list node into symbol

From: Ian Rogers
Date: Mon Mar 20 2023 - 13:00:36 EST

Next message: Jim Mattson: "Re: [PATCH 0/3] KVM: support the cpu feature FLUSH_L1D"
Previous message: Nishanth Menon: "[PATCH V3 1/2] arm64: dts: ti: k3-am62-wakeup: Introduce RTC node"
In reply to: Arnaldo Carvalho de Melo: "Re: [PATCH v4 06/22] perf map: Move map list node into symbol"
Next in thread: Ian Rogers: "[PATCH v4 07/22] perf maps: Remove rb_node from struct map"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Mon, Mar 20, 2023 at 8:48 AM Arnaldo Carvalho de Melo
<acme@xxxxxxxxxx> wrote:
>
> Em Sun, Mar 19, 2023 at 08:37:54PM -0700, Ian Rogers escreveu:
> > Using a perf map as a list node is only done in symbol. Move the
> > list_node struct into symbol as a single pointer to the map. This
> > makes reference count behavior more obvious and easy to check.
> >
> > Signed-off-by: Ian Rogers <irogers@xxxxxxxxxx>
> > ---
> > tools/perf/util/map.h | 5 +--
> > tools/perf/util/symbol.c | 90 ++++++++++++++++++++++++++--------------
> > 2 files changed, 60 insertions(+), 35 deletions(-)
> >
> > diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h
> > index 3dcfe06db6b3..2879cae05ee0 100644
> > --- a/tools/perf/util/map.h
> > +++ b/tools/perf/util/map.h
> > @@ -16,10 +16,7 @@ struct maps;
> > struct machine;
> >
> > struct map {
> > - union {
> > - struct rb_node rb_node;
> > - struct list_head node;
> > - };
> > + struct rb_node rb_node;
> > u64 start;
> > u64 end;
> > bool erange_warned:1;
> > diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
> > index a458aa8b87bb..2676a163e237 100644
> > --- a/tools/perf/util/symbol.c
> > +++ b/tools/perf/util/symbol.c
> > @@ -48,6 +48,11 @@ static bool symbol__is_idle(const char *name);
> > int vmlinux_path__nr_entries;
> > char **vmlinux_path;
> >
> > +struct map_list_node {
> > + struct list_head node;
> > + struct map *map;
> > +};
> > +
> > struct symbol_conf symbol_conf = {
> > .nanosecs = false,
> > .use_modules = true,
> > @@ -1219,16 +1224,22 @@ struct kcore_mapfn_data {
> > static int kcore_mapfn(u64 start, u64 len, u64 pgoff, void *data)
> > {
> > struct kcore_mapfn_data *md = data;
> > - struct map *map;
> > + struct map_list_node *list_node;
> > +
> > + list_node = malloc(sizeof(*list_node));
> > + if (list_node == NULL)
> > + return -ENOMEM;
>
> So this is open coding map_list_node__new(), can we have it defined
> right after the struct map_list_node definition?
>
> I've applied the 1-5 patches in this series to my local branch, testing
> now.

Thanks Arnaldo. Done for v5 as a wrapper of malloc as each call site
modifies the map in unique ways.

Ian

> - Arnaldo
>
> >
> > - map = map__new2(start, md->dso);
> > - if (map == NULL)
> > + list_node->map = map__new2(start, md->dso);
> > + if (list_node->map == NULL) {
> > + free(list_node);
> > return -ENOMEM;
> > + }
> >
> > - map->end = map->start + len;
> > - map->pgoff = pgoff;
> > + list_node->map->end = list_node->map->start + len;
> > + list_node->map->pgoff = pgoff;
> >
> > - list_add(&map->node, &md->maps);
> > + list_add(&list_node->node, &md->maps);
> >
> > return 0;
> > }
> > @@ -1264,12 +1275,19 @@ int maps__merge_in(struct maps *kmaps, struct map *new_map)
> > * |new.............| -> |new..| |new..|
> > * |old....| -> |old....|
> > */
> > - struct map *m = map__clone(new_map);
> > + struct map_list_node *m;
> >
> > + m = malloc(sizeof(*m));
> > if (!m)
> > return -ENOMEM;
> >
> > - m->end = old_map->start;
> > + m->map = map__clone(new_map);
> > + if (!m->map) {
> > + free(m);
> > + return -ENOMEM;
> > + }
> > +
> > + m->map->end = old_map->start;
> > list_add_tail(&m->node, &merged);
> > new_map->pgoff += old_map->end - new_map->start;
> > new_map->start = old_map->end;
> > @@ -1299,10 +1317,13 @@ int maps__merge_in(struct maps *kmaps, struct map *new_map)
> > }
> >
> > while (!list_empty(&merged)) {
> > - old_map = list_entry(merged.next, struct map, node);
> > - list_del_init(&old_map->node);
> > - maps__insert(kmaps, old_map);
> > - map__put(old_map);
> > + struct map_list_node *old_node;
> > +
> > + old_node = list_entry(merged.next, struct map_list_node, node);
> > + list_del_init(&old_node->node);
> > + maps__insert(kmaps, old_node->map);
> > + map__put(old_node->map);
> > + free(old_node);
> > }
> >
> > if (new_map) {
> > @@ -1317,7 +1338,7 @@ static int dso__load_kcore(struct dso *dso, struct map *map,
> > {
> > struct maps *kmaps = map__kmaps(map);
> > struct kcore_mapfn_data md;
> > - struct map *old_map, *new_map, *replacement_map = NULL, *next;
> > + struct map *old_map, *replacement_map = NULL, *next;
> > struct machine *machine;
> > bool is_64_bit;
> > int err, fd;
> > @@ -1378,11 +1399,12 @@ static int dso__load_kcore(struct dso *dso, struct map *map,
> > /* Find the kernel map using the '_stext' symbol */
> > if (!kallsyms__get_function_start(kallsyms_filename, "_stext", &stext)) {
> > u64 replacement_size = 0;
> > + struct map_list_node *new_node;
> >
> > - list_for_each_entry(new_map, &md.maps, node) {
> > - u64 new_size = new_map->end - new_map->start;
> > + list_for_each_entry(new_node, &md.maps, node) {
> > + u64 new_size = new_node->map->end - new_node->map->start;
> >
> > - if (!(stext >= new_map->start && stext < new_map->end))
> > + if (!(stext >= new_node->map->start && stext < new_node->map->end))
> > continue;
> >
> > /*
> > @@ -1392,40 +1414,43 @@ static int dso__load_kcore(struct dso *dso, struct map *map,
> > * falls within more than one in the list.
> > */
> > if (!replacement_map || new_size < replacement_size) {
> > - replacement_map = new_map;
> > + replacement_map = new_node->map;
> > replacement_size = new_size;
> > }
> > }
> > }
> >
> > if (!replacement_map)
> > - replacement_map = list_entry(md.maps.next, struct map, node);
> > + replacement_map = list_entry(md.maps.next, struct map_list_node, node)->map;
> >
> > /* Add new maps */
> > while (!list_empty(&md.maps)) {
> > - new_map = list_entry(md.maps.next, struct map, node);
> > - list_del_init(&new_map->node);
> > - if (new_map == replacement_map) {
> > - map->start = new_map->start;
> > - map->end = new_map->end;
> > - map->pgoff = new_map->pgoff;
> > - map->map_ip = new_map->map_ip;
> > - map->unmap_ip = new_map->unmap_ip;
> > + struct map_list_node *new_node;
> > +
> > + new_node = list_entry(md.maps.next, struct map_list_node, node);
> > + list_del_init(&new_node->node);
> > + if (new_node->map == replacement_map) {
> > + map->start = new_node->map->start;
> > + map->end = new_node->map->end;
> > + map->pgoff = new_node->map->pgoff;
> > + map->map_ip = new_node->map->map_ip;
> > + map->unmap_ip = new_node->map->unmap_ip;
> > /* Ensure maps are correctly ordered */
> > map__get(map);
> > maps__remove(kmaps, map);
> > maps__insert(kmaps, map);
> > map__put(map);
> > - map__put(new_map);
> > + map__put(new_node->map);
> > } else {
> > /*
> > * Merge kcore map into existing maps,
> > * and ensure that current maps (eBPF)
> > * stay intact.
> > */
> > - if (maps__merge_in(kmaps, new_map))
> > + if (maps__merge_in(kmaps, new_node->map))
> > goto out_err;
> > }
> > + free(new_node);
> > }
> >
> > if (machine__is(machine, "x86_64")) {
> > @@ -1462,9 +1487,12 @@ static int dso__load_kcore(struct dso *dso, struct map *map,
> >
> > out_err:
> > while (!list_empty(&md.maps)) {
> > - map = list_entry(md.maps.next, struct map, node);
> > - list_del_init(&map->node);
> > - map__put(map);
> > + struct map_list_node *list_node;
> > +
> > + list_node = list_entry(md.maps.next, struct map_list_node, node);
> > + list_del_init(&list_node->node);
> > + map__put(list_node->map);
> > + free(list_node);
> > }
> > close(fd);
> > return -EINVAL;
> > --
> > 2.40.0.rc1.284.g88254d51c5-goog
> >
>
> --
>
> - Arnaldo

Next message: Jim Mattson: "Re: [PATCH 0/3] KVM: support the cpu feature FLUSH_L1D"
Previous message: Nishanth Menon: "[PATCH V3 1/2] arm64: dts: ti: k3-am62-wakeup: Introduce RTC node"
In reply to: Arnaldo Carvalho de Melo: "Re: [PATCH v4 06/22] perf map: Move map list node into symbol"
Next in thread: Ian Rogers: "[PATCH v4 07/22] perf maps: Remove rb_node from struct map"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]