From: David Howells Date: Wed, 09 Aug 2017 09:33:53 +0000 Subject: Re: [PATCH 3/3] net: rxrpc: Replace time_t type with time64_t type Message-Id: <6767.1502271233@warthog.procyon.org.uk> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit List-Id: References: <8ac57c96bf5a0695ecc67fd230440b0b9d15740f.1502246502.git.baolin.wang@linaro.org> In-Reply-To: To: linux-security-module@vger.kernel.org Arnd Bergmann wrote: > > @@ -533,8 +536,9 @@ static int rxrpc_preparse_xdr_rxk5(struct key_preparsed_payload *prep, > > pptoken = &(*pptoken)->next) > > continue; > > *pptoken = token; > > - if (token->kad->expiry < prep->expiry) > > - prep->expiry = token->kad->expiry; > ... > > I'm still slightly puzzled by what this function does: it does have four > timestamps (authtime, starttime, endtime, renew_till) that are all > transferred as 64-bit values and won't expire, but then it also uses the > 32-bit expiry field in rxrpc_key_token->kad->expiry instead of the 64-bit > rxrpc_key_token->k5 fields. Good catch. This is a cut'n'paste error. It should be using token->k5->expiry here not token->kad->expiry. > This appears to overlay the first 32 bits of the > rxrpc_key_token->k5->starttime field, which is also a time value on > little-endian architectures by chance, but I would assume that it's always > in the past, meaning the keys would already be expired. Yeah - I'm not sure why it works. David