Hi Alexander,
On 5/10/26 09:47, Solar Designer wrote:
On Sat, May 09, 2026 at 08:18:49PM +0200, Sebastian Pipping wrote:
just a quick note that uriparser 1.0.2 released today is fixing
vulnerabilities CVE-2026-44927 and CVE-2026-44928.
Thanks, but let's please be including vulnerability descriptions right
in the postings. Also, when it's one vulnerability, its title should be
in the Subject line. When it's more than one, then if there's a way to
group them e.g. by category or severity, that could go into the Subject.
Some key links are:
- The change log of release 1.0.2
https://github.com/uriparser/uriparser/blob/uriparser-1.0.2/ChangeLog
This says:
SECURITY >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
* Fixed: [CVE-2026-44927]
Stop truncating `ptrdiff_t` to `int`
Thanks for the report to Ilia Alshanetsky and Joshua W. Windle!
(GitHub #304)
* Fixed: [CVE-2026-44928]
Fix `EqualsUri` with regard to `.absolutePath`
Thanks for the report to Ilia Alshanetsky! (GitHub #305)
SECURITY >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
* Fixed: Fix OOM related memory leak in `CopyUriMm`
Thanks for the report to Ilia Alshanetsky! (GitHub #306)
- The fixing pull requests
- https://github.com/uriparser/uriparser/pull/304
- https://github.com/uriparser/uriparser/pull/305
- The official CVE metadata
- https://nvd.nist.gov/vuln/detail/CVE-2026-44927
- https://nvd.nist.gov/vuln/detail/CVE-2026-44928
These only say a little:
CVE-2026-44927: In uriparser before 1.0.2, there is pointer difference
truncation to int in various places.
CVE-2026-44928: In uriparser before 1.0.2, the function family EqualsUri
can misclassify two unequal URIs as equal.
For CVE-2026-44927, it could help to clarify actual security exposure
and impact.
I understand.
My understanding of the impact of the ptrdiff_t truncation
(CVE-2026-44927) was "DoS or more", and I focussed on fixing this in the
about 15 places rather than analyzing the impact in more detail.
Ilia and Joshua considered impact more than me, and I'm BCC'ing them
so they can reply to this mail directly if they like, please do not feel
pushed.
Let me add that the public CVSS score input for CVE-2026-44927 is
(again) mistaken; none of these are correct:
- Attack vector: Local
- Attack complexity: High
- Availability: None
(Seen at https://github.com/advisories/ghsa-gmxg-5w57-j63q just now.)
Best
Sebastian
