Am Montag, 5. Januar 2015, 21:51:06 schrieb Herbert Xu: Hi Herbert, > On Mon, Jan 05, 2015 at 11:46:50AM +0100, Stephan Mueller wrote: > > The need for that check lies in aead_recvmsg: > > /* > > > > * first chunk of input is AD -- one scatterlist entry is one > > page, > > * and we process only one scatterlist, the maximum size of AD is > > * one page > > */ > > > > sg_init_table(&assoc, 1); > > sg_set_page(&assoc, sg_page(sg), ctx->aead_assoclen, sg->offset); > > aead_request_set_assoc(&ctx->aead_req, &assoc, > > ctx->aead_assoclen); > > > > There you see that I only create an sg table with one entry for the AD. If > > we would allow an arbitrary AD size, I would see the need of a for loop > > in addition to the one directly beneath this AD scatterlist setting: one > > for identifying how many sg entries I need to allocate and one for the > > actual assignment. > > > > As I felt that one page should be sufficient for the AD, I wanted to avoid > > the extra overhead for another for loop. > > Please remove the limit as otherwise we would never be able to > add support for this in a future kernel as appliations won't be > able to rely on it. > > There is no such limit in the kernel interface and we shouldn't > be adding one here. One question: are you aware of an existing mechanism to split one scatterlist into two at a given offset? > > Cheers, -- Ciao Stephan -- To unsubscribe from this list: send the line "unsubscribe linux-crypto" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
