Re: [PATCH] usertools/pmdinfo: fix search for PMD info string

[Thierry Herbelot <thierry.herbelot@6wind.com>]

On 2/17/26 18:59, Bruce Richardson wrote:
> On Tue, Feb 17, 2026 at 06:53:20PM +0100, David Marchand wrote:
>> On Tue, 17 Feb 2026 at 13:49, Bruce Richardson
>> <bruce.richardson@intel.com> wrote:
>>>
>>> The PMD_INFO_STRING constant string is not guaranteed to appear in the
>>> output binary immediately after an unprintable character. Because of
>>> this, in some cases the information for a driver could be missed by the
>>> PMD info script as it only checks for the prefix at the start of strings
>>> that it finds. Change the script to use "s.find()" rather than
>>> "s.startswith()" to fix this issue.
>>>
>>> Fixes: 0ce3cf4afd04 ("usertools/pmdinfo: rewrite simpler script")
>>> Cc: stable@dpdk.org
>>>
>>> Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
>>
>> Surprising we hit this issue only now.
> 
> Agreed. Two thoughts here:
> 1. are the compilers normally arranging the data in ways that nearly
> guarantee that the strings are not prefixed by other printable data?
> 2. if we were missing a couple of drivers from dpdk-pmdinfo, would it
> really be noticed? Even in the reported case, there were actually 3 missing
> driver listings, but only one was actually noticed.
> 
> I'd suspect #2 myself. :-)
> 
>> Acked-by: David Marchand <david.marchand@redhat.com>
>>
>> It would be good to hear back from the initial reporter.
>>
> Agreed. I'd like independent confirmation that this fixes the issue on RHEL
> 9 as reported.

Hello Bruce,

Sorry for the late response.
I confirm your patch does fix the original issue:

$ head -2 /etc/os-release
NAME="Red Hat Enterprise Linux"
VERSION="9.7 (Plow)"
$ meson setup build
$ ninja -C build
...
$ ./usertools/dpdk-pmdinfo.py build/app/dpdk-testpmd|grep ixgbe
     "name": "net_ixgbe_vf",
     "name": "net_ixgbe",

Thanks !

	Best regards

	Thierry

> 
> /Bruce

-- 
Thierry Herbelot