Your message dated Tue, 03 May 2022 17:14:03 +0530
with message-id <[email protected]>
and subject line Re: bpfcc-tools: biosnoop fails to print the SIZE field
correctly on bullseye
has caused the Debian Bug report #996961,
regarding bpfcc-tools: biosnoop fails to print the SIZE field correctly on
bullseye
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)
--
996961: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=996961
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Package: bpfcc-tools
Version: 0.18.0+ds-2
Severity: normal
Dear Maintainer,
Pretty simple - it looks like the internal field that biosnoop-bpfcc reads
to get the size of the IO started being cleared on IO completion at some point,
so it prints as always 0 now.
Upstream fixed this by stashing the size and timestamp before the IO starts;
since
that's really the only delta between the two, I'd probably just take that
wholesale (the
patch attached is a delta between what's currently in 0.18.0 and git master).
(Don't mind the debsums error, I went and tinkered with my copy to try a couple
of
solutions; it's broken on unmodified biosnoop.)
- Rich
-- System Information:
Debian Release: 11.1
APT prefers stable-updates
APT policy: (1000, 'stable-updates'), (1000, 'stable-security'), (1000,
'stable'), (901, 'proposed-updates'), (900, 'oldstable-debug'), (900,
'testing'), (800, 'unstable-debug'), (500, 'stable-debug'), (500,
'proposed-updates-debug'), (500, 'oldstable-proposed-updates-debug'), (1,
'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386
Kernel: Linux 5.10.0-9-amd64 (SMP w/16 CPU threads)
Kernel taint flags: TAINT_PROPRIETARY_MODULE, TAINT_OOT_MODULE,
TAINT_UNSIGNED_MODULE
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8), LANGUAGE not set
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled
Versions of packages bpfcc-tools depends on:
ii python3 3.9.2-3
ii python3-bpfcc 0.18.0+ds-2
ii python3-netaddr 0.7.19-5
bpfcc-tools recommends no packages.
bpfcc-tools suggests no packages.
-- no debconf information
-- debsums errors found:
debsums: changed file /usr/sbin/biosnoop-bpfcc (from bpfcc-tools package)
--- tools/biosnoop.py.old 2021-10-21 09:38:53.626276801 -0400
+++ tools/biosnoop.py 2021-10-21 09:39:03.354295218 -0400
@@ -39,6 +39,12 @@
#include <uapi/linux/ptrace.h>
#include <linux/blkdev.h>
+// for saving the timestamp and __data_len of each request
+struct start_req_t {
+ u64 ts;
+ u64 data_len;
+};
+
struct val_t {
u64 ts;
u32 pid;
@@ -57,7 +63,7 @@
char name[TASK_COMM_LEN];
};
-BPF_HASH(start, struct request *);
+BPF_HASH(start, struct request *, struct start_req_t);
BPF_HASH(infobyreq, struct request *, struct val_t);
BPF_PERF_OUTPUT(events);
@@ -80,42 +86,43 @@
// time block I/O
int trace_req_start(struct pt_regs *ctx, struct request *req)
{
- u64 ts;
- ts = bpf_ktime_get_ns();
- start.update(&req, &ts);
+ struct start_req_t start_req = {
+ .ts = bpf_ktime_get_ns(),
+ .data_len = req->__data_len
+ };
+ start.update(&req, &start_req);
return 0;
}
// output
int trace_req_completion(struct pt_regs *ctx, struct request *req)
{
- u64 *tsp;
+ struct start_req_t *startp;
struct val_t *valp;
struct data_t data = {};
u64 ts;
// fetch timestamp and calculate delta
- tsp = start.lookup(&req);
- if (tsp == 0) {
+ startp = start.lookup(&req);
+ if (startp == 0) {
// missed tracing issue
return 0;
}
ts = bpf_ktime_get_ns();
- data.delta = ts - *tsp;
+ data.delta = ts - startp->ts;
data.ts = ts / 1000;
data.qdelta = 0;
valp = infobyreq.lookup(&req);
+ data.len = startp->data_len;
if (valp == 0) {
- data.len = req->__data_len;
data.name[0] = '?';
data.name[1] = 0;
} else {
if (##QUEUE##) {
- data.qdelta = *tsp - valp->ts;
+ data.qdelta = startp->ts - valp->ts;
}
data.pid = valp->pid;
- data.len = req->__data_len;
data.sector = req->__sector;
bpf_probe_read_kernel(&data.name, sizeof(data.name), valp->name);
struct gendisk *rq_disk = req->rq_disk;
--- End Message ---
--- Begin Message ---
Control: fixed -1 0.24.0+ds-1
Hi Rich,
Sorry for the delayed response. This is no longer an issue in latest
version of bpfcc-tools. Hence closing the bug.
Thanks and Regards,
Vasudev
--- End Message ---
