netdev - Re: [RFC PATCH 1/2] net: Add sysfs files for threaded NAPI.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Wed, 24 May 2023 16:43:44 +0200
From: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
To: Eric Dumazet <edumazet@...gle.com>
Cc: netdev@...r.kernel.org, Jakub Kicinski <kuba@...nel.org>,
	Kurt Kanzenbach <kurt.kanzenbach@...utronix.de>,
	Paolo Abeni <pabeni@...hat.com>,
	Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [RFC PATCH 1/2] net: Add sysfs files for threaded NAPI.

On 2023-05-24 15:53:27 [+0200], Eric Dumazet wrote:
> How is interface rename handled ?
> 
> root@...mazet1:~# ip link show dev dummy0
> 4: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN mode
> DEFAULT group default qlen 1000
>     link/ether f2:38:20:69:b4:ca brd ff:ff:ff:ff:ff:ff
> root@...mazet1:~# ip link set dummy0 name new-name
> root@...mazet1:~# ip link show dev dummy0
> Device "dummy0" does not exist.
> root@...mazet1:~# ip link show dev new-name
> 4: new-name: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN mode
> DEFAULT group default qlen 1000
>     link/ether f2:38:20:69:b4:ca brd ff:ff:ff:ff:ff:ff

Not sure I understood the question. But after
	ip link set eno0 name newdev

I have
| root@box:/sys/class/net/newdev# ps uax | grep napi
| root        2246  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8200]
| root        2247  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8199]
| root        2248  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8198]
| root        2249  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8197]
| root        2250  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8196]
| root        2251  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8195]
| root        2252  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8194]
| root        2253  0.0  0.0      0     0 ?        S    12:04   0:00 [napi/eno0-8193]
|
| root@box:/sys/class/net/newdev# ls -lh napi
| total 0
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-0
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-1
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-2
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-3
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-4
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-5
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-6
| drwxr-xr-x 2 root root 0 May 24 12:05 eno0-TxRx-7

NAPI was not freed/ allocated again. For some reason the interface went
down during the rename. I suspect the fancy network manager because it
does not know the interface. It did noto happen again after a further
rename of the renamed device and it remained up.

A link up request resulted in requesting the interrupts again and so the
names of the IRQ-threads contain now "newdev" in /proc/interrupts and
the relevant interrupt thread. This isn't the case if I rename it again
since the device isn't going down.

If I reconfigure the queues
	ethtool -L newdev combined 6 other 1

which drops all NAPI devices and asks for new then the NAPI threads are
displayed properly (in ps) but the sysfs entries in the napi folder are
napi_id-$id so I probably missed a spot in the igb driver. And reverting
back to the original 8 channels does not change the fact that I still
see napi_id-$id. So details… But the interrupt number and so on match :)

> Thanks.

Sebastian