From: Jacob Carlborg <doob@me.com>
To: dtrace@lists.linux.dev
Cc: Jacob Carlborg <doob@me.com>
Subject: [PATCH 2/3] libdtrace: derive io provider dev_name from the device model
Date: Mon, 20 Jul 2026 15:30:59 +0200 [thread overview]
Message-ID: <20260720133100.77573-3-doob@me.com> (raw)
In-Reply-To: <20260720133100.77573-1-doob@me.com>
Without this fix, running any DTrace commands, like `dtrace -l`,
results in:
```
dtrace: invalid probe specifier :::: "/usr/lib64/dtrace/6.12.0/io.d",
line 134: failed to resolve vmlinux`major_names: Unknown symbol name
```
The issue is that on newer kernels the `major_names` symbols is not
available in `kallsyms` anymore. For example, running `grep -w
major_names /proc/kallsyms` on a newer kernel shows no results.
Running the same command on an older kernel gives the following
result:
```
$ grep -w major_names /proc/kallsyms
0000000000000000 b major_names
```
The fix is to stop depending on the `major_names` symbol and instead
derive `dev_name` from the device model (the driver name), mirroring
the `struct buffer_head` translator.
Signed-off-by: Jacob Carlborg <doob@me.com>
---
libdtrace/io.d.in | 17 ++++++++++++++---
1 file changed, 14 insertions(+), 3 deletions(-)
diff --git a/libdtrace/io.d.in b/libdtrace/io.d.in
index 5ea356ee..e96a5b23 100644
--- a/libdtrace/io.d.in
+++ b/libdtrace/io.d.in
@@ -124,16 +124,27 @@ translator devinfo_t < struct buffer_head *B > {
);
};
+define_for_kernel([[__part0_dev]], [[(m4_kver(5,11,0), [[part0->bd_device]])]], [[part0.__dev]])
#pragma D binding "1.6.3" translator
translator devinfo_t < struct bio *B > {
dev_major = B->__disk_chk == NULL ? 0 : getmajor(B->__bio_part_dev);
dev_minor = B->__disk_chk == NULL ? 0 : getminor(B->__bio_part_dev);
dev_instance = 0;
+ /*
+ * Derive the driver name from the device model rather than the kernel's
+ * `major_names` array: that symbol became static and is no longer
+ * resolvable via kallsyms on newer kernels. This mirrors how the
+ * struct buffer_head devinfo translator above obtains dev_name.
+ */
dev_name = B->__disk_chk == NULL
? "nfs"
- : stringof(((struct blk_major_name **)vmlinux`major_names)[
- getmajor(B->__bio_part_dev) % 255
- ]->name);
+ : B->__disk->__part0_dev.parent
+ ? B->__disk->__part0_dev.parent->driver->name
+ ? stringof(B->__disk->__part0_dev.parent->driver->name)
+ : "<none>"
+ : B->__disk->__part0_dev.driver->name
+ ? stringof(B->__disk->__part0_dev.driver->name)
+ : "<none>";
dev_statname = B->__disk_chk == NULL ? "nfs" :
(B->__bio_partno) == 0 ? stringof(B->__disk->disk_name) :
strjoin(stringof(B->__disk->disk_name), lltostr(B->__bio_partno));
--
2.50.1 (Apple Git-155)
next prev parent reply other threads:[~2026-07-20 13:31 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-07-20 13:30 [PATCH 0/3] Fix DTrace on modern kernels and in containers Jacob Carlborg
2026-07-20 13:30 ` [PATCH 1/3] libdtrace: resolve BTF type tags when importing BTF into CTF Jacob Carlborg
2026-07-20 13:30 ` Jacob Carlborg [this message]
2026-07-20 13:31 ` [PATCH 3/3] libdtrace: match BEGIN/END PID in the tracer's own PID namespace Jacob Carlborg
2026-07-20 16:33 ` [PATCH 0/3] Fix DTrace on modern kernels and in containers Kris Van Hees
2026-07-21 6:27 ` Jacob Carlborg
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=20260720133100.77573-3-doob@me.com \
--to=doob@me.com \
--cc=dtrace@lists.linux.dev \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox