From mboxrd@z Thu Jan 1 00:00:00 1970 Content-Type: multipart/mixed; boundary="===============7919235672037531072==" MIME-Version: 1.0 From: Florian Westphal To: mptcp at lists.01.org Subject: Re: [MPTCP] [PATCH] mptcp: fix mptcp socket cleanup Date: Fri, 30 Aug 2019 14:29:21 +0200 Message-ID: <20190830122921.GP20113@breakpoint.cc> In-Reply-To: e88e1256082439491fbfb0b92e254a692b32af7f.1567167420.git.pabeni@redhat.com X-Status: X-Keywords: X-UID: 1768 --===============7919235672037531072== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Paolo Abeni wrote: > + /* when assiging the token to the msk, we need to held > + * an additional reference, will be released at msk token > + * destruction time > + */ Unrelated to your patch, but I think this is wrong. We should have implicit reference for the token. Taking a reference makes no sense because the token lifetime is limited by the mptcp socket -- if the mptcp socket is closing, we remove the token too. token_release() function makes no sense to me -- afaics all callers have a pointer to the msk they want to put the reference for. I'll have a look into cleaning that up after respinning a squashable version of the token refactoring. --===============7919235672037531072==--