diff options
| author | Tejun Heo <tj@kernel.org> | 2026-09-14 22:12:34 -1000 |
|---|---|---|
| committer | Tejun Heo <tj@kernel.org> | 2026-09-15 06:57:20 -1000 |
| commit | a9e3760b0838299649c0d57cca44daaf40ba3c33 (patch) | |
| tree | 95807ff0e511c05602ef34808cfc660303309057 /tools/writeback | |
| download | linux-stable-a9e3760b0838299649c0d57cca44daaf40ba3c33.tar.gz linux-stable-a9e3760b0838299649c0d57cca44daaf40ba3c33.zip | |
sched_ext: Maintain an online cid mask in the scheduler arenagrafted
Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.
Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.
The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.
v2: Reworded the getter kerneldoc (Andrea Righi).
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Diffstat (limited to 'tools/writeback')
| -rw-r--r-- | tools/writeback/wb_monitor.py | 172 |
1 files changed, 172 insertions, 0 deletions
diff --git a/tools/writeback/wb_monitor.py b/tools/writeback/wb_monitor.py new file mode 100644 index 000000000..5e3591f1f --- /dev/null +++ b/tools/writeback/wb_monitor.py @@ -0,0 +1,172 @@ +#!/usr/bin/env drgn +# +# Copyright (C) 2024 Kemeng Shi <shikemeng@huaweicloud.com> +# Copyright (C) 2024 Huawei Inc + +desc = """ +This is a drgn script based on wq_monitor.py to monitor writeback info on +backing dev. For more info on drgn, visit https://github.com/osandov/drgn. + + writeback(kB) Amount of dirty pages are currently being written back to + disk. + + reclaimable(kB) Amount of pages are currently reclaimable. + + dirtied(kB) Amount of pages have been dirtied. + + wrttien(kB) Amount of dirty pages have been written back to disk. + + avg_wb(kBps) Smoothly estimated write bandwidth of writing dirty pages + back to disk. +""" + +import signal +import re +import time +import json + +import drgn +from drgn.helpers.linux.list import list_for_each_entry + +import argparse +parser = argparse.ArgumentParser(description=desc, + formatter_class=argparse.RawTextHelpFormatter) +parser.add_argument('bdi', metavar='REGEX', nargs='*', + help='Target backing device name patterns (all if empty)') +parser.add_argument('-i', '--interval', metavar='SECS', type=float, default=1, + help='Monitoring interval (0 to print once and exit)') +parser.add_argument('-j', '--json', action='store_true', + help='Output in json') +parser.add_argument('-c', '--cgroup', action='store_true', + help='show writeback of bdi in cgroup') +args = parser.parse_args() + +bdi_list = prog['bdi_list'] + +WB_RECLAIMABLE = prog['WB_RECLAIMABLE'] +WB_WRITEBACK = prog['WB_WRITEBACK'] +WB_DIRTIED = prog['WB_DIRTIED'] +WB_WRITTEN = prog['WB_WRITTEN'] +NR_WB_STAT_ITEMS = prog['NR_WB_STAT_ITEMS'] + +PAGE_SHIFT = prog['PAGE_SHIFT'] + +def K(x): + return x << (PAGE_SHIFT - 10) + +class Stats: + def dict(self, now): + return { 'timestamp' : now, + 'name' : self.name, + 'writeback' : self.stats[WB_WRITEBACK], + 'reclaimable' : self.stats[WB_RECLAIMABLE], + 'dirtied' : self.stats[WB_DIRTIED], + 'written' : self.stats[WB_WRITTEN], + 'avg_wb' : self.avg_bw, } + + def table_header_str(): + return f'{"":>16} {"writeback":>10} {"reclaimable":>12} ' \ + f'{"dirtied":>9} {"written":>9} {"avg_bw":>9}' + + def table_row_str(self): + out = f'{self.name[-16:]:16} ' \ + f'{self.stats[WB_WRITEBACK]:10} ' \ + f'{self.stats[WB_RECLAIMABLE]:12} ' \ + f'{self.stats[WB_DIRTIED]:9} ' \ + f'{self.stats[WB_WRITTEN]:9} ' \ + f'{self.avg_bw:9} ' + return out + + def show_header(): + if Stats.table_fmt: + print() + print(Stats.table_header_str()) + + def show_stats(self): + if Stats.table_fmt: + print(self.table_row_str()) + else: + print(self.dict(Stats.now)) + +class WbStats(Stats): + def __init__(self, wb): + bdi_name = wb.bdi.dev_name.string_().decode() + # avoid to use bdi.wb.memcg_css which is only defined when + # CONFIG_CGROUP_WRITEBACK is enabled + if wb == wb.bdi.wb.address_of_(): + ino = "1" + else: + ino = str(wb.memcg_css.cgroup.kn.id.value_()) + self.name = bdi_name + '_' + ino + + self.stats = [0] * NR_WB_STAT_ITEMS + for i in range(NR_WB_STAT_ITEMS): + if wb.stat[i].count >= 0: + self.stats[i] = int(K(wb.stat[i].count)) + else: + self.stats[i] = 0 + + self.avg_bw = int(K(wb.avg_write_bandwidth)) + +class BdiStats(Stats): + def __init__(self, bdi): + self.name = bdi.dev_name.string_().decode() + self.stats = [0] * NR_WB_STAT_ITEMS + self.avg_bw = 0 + + def collectStats(self, wb_stats): + for i in range(NR_WB_STAT_ITEMS): + self.stats[i] += wb_stats.stats[i] + + self.avg_bw += wb_stats.avg_bw + +exit_req = False + +def sigint_handler(signr, frame): + global exit_req + exit_req = True + +def main(): + # handle args + Stats.table_fmt = not args.json + interval = args.interval + cgroup = args.cgroup + + re_str = None + if args.bdi: + for r in args.bdi: + if re_str is None: + re_str = r + else: + re_str += '|' + r + + filter_re = re.compile(re_str) if re_str else None + + # monitoring loop + signal.signal(signal.SIGINT, sigint_handler) + + while not exit_req: + Stats.now = time.time() + + Stats.show_header() + for bdi in list_for_each_entry('struct backing_dev_info', bdi_list.address_of_(), 'bdi_list'): + bdi_stats = BdiStats(bdi) + if filter_re and not filter_re.search(bdi_stats.name): + continue + + for wb in list_for_each_entry('struct bdi_writeback', bdi.wb_list.address_of_(), 'bdi_node'): + wb_stats = WbStats(wb) + bdi_stats.collectStats(wb_stats) + if cgroup: + wb_stats.show_stats() + + bdi_stats.show_stats() + if cgroup and Stats.table_fmt: + print() + + if interval == 0: + break + time.sleep(interval) + +if __name__ == "__main__": + main() |
