Bug#1110607: libsoup3: CVE-2025-8197: out-of-bounds read in soup_header_name_to_string()

[Simon McVittie]

Control: retitle -1 libsoup3: CVE-2025-8197: out-of-bounds read in soup_header_name_to_string()
Control: forwarded -1 https://gitlab.gnome.org/GNOME/libsoup/-/issues/465
Control: tags -1 + moreinfo

On Sat, 09 Aug 2025 at 11:51:29 +0200, Salvatore Bonaccorso wrote:
>CVE-2025-8197[0]:
>| A global buffer overflow vulnerability was found in the
>| soup_header_name_to_string function in Libsoup. The
>| `soup_header_name_to_string` function does not validate the `name`
>| parameter passed in, and directly accesses
>| `soup_header_name_strings[name]`. The value of `name` is
>| controllable, when `name` exceeds the index range of
>| `soup_headr_name_string`, it will cause an out-of-bounds access.
>
>There is only one reference to the Red Hat Bugzilla, and have not
>found an upstream issue or a merge request referencing the CVE, can
>you double-check?

I couldn't find an upstream issue either, so I opened one. It is not 
clear to me that this is a genuine vulnerability. I think this needs 
clarification from the CNA that allocated a CVE ID for this (which 
appears to be Red Hat).

At best, it's a misleading description of a genuine vulnerability, 
something like "by doing foo, bar and baz, an attacker can trick libsoup 
into calling soup_header_name_to_string() with an invalid argument, 
which crashes" - but if that's the case, then I think it's the caller of 
soup_header_name_to_string() that should be fixed, and not 
soup_header_name_to_string() itself.

     smcv