Re: [RFC][PATCH] Make cryptoapi non-optional?

[Adam J. Richter]

At 2003-08-16 15:51:10, Matt Mackall wrote:
>The
>current code uses the stack (though currently rather a lot of it),
>which lets it be fully re-entrant. Not an option with cryptoapi.

	I posted a patch a while ago on one of the linux crypto
mailing lists that defined these routines to support allocating
crypto_tfm's on the stack:

int crypto_tfm_alloc_size(struct crypto_alg *alg, u32 tfm_flags);
int crypto_tfm_init(struct crypto_tfm *tfm, struct crypto_alg *alg,
                    u32 tfm_flags);

	The patch also created crypto_alg_{get,put} routines so that
crypto_tfm's could be created quickly without having to look up and
release references to crypto_alg's.

	I have a version of cryptoloop.c that uses these changes so
write operations on device-backed loop devices will not corrupt data
(I beileve that in the stock 2.6.0-test3 this data corruption can
happen, although I've never observed it).  The changes should also
facilitate a future version of loop.c that may be able to have
encryptions or decryption for the same /dev/loop/ device running on
multiple CPU's simultaneously.

	If there is interest, I can assemble a new version of the
patch for 2.6.0-test3.

Adam J. Richter     __     ______________   575 Oroville Road
adam@xxxxxxxxxxxxx     \ /                  Miplitas, California 95035
+1 408 309-6081         | g g d r a s i l   United States of America
                         "Free Software For The Rest Of Us."
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/