Re: [Cryptodev-linux-devel] [PATCH] RFC: third working version of the async ops

Hey, On Wed, Oct 20, 2010 at 12:35:51PM +0200, Nikos Mavrogiannopoulos wrote: On Wed, Oct 13, 2010 at 5:17 PM, Phil Sutter phil.sut...@viprinet.com wrote: Still ugly as hell (and therefore TODO): - having to pass task_struct and mm_struct from the calling process  around - need to copy

[Cryptodev-linux-devel] [PATCH 5/8] implement asynchronous operation mode

In order to submit a job for asynchronous completion, one uses the ioctl CIOCASYNCCRYPT, which is syntactically equal to CIOCCRYPT. But, the former will return immediately, the result has to be fetched later using CIOCASYNCFETCH (which will always return the first completed job, asking for a

[Cryptodev-linux-devel] [PATCH 2/8] keep a copy of the IV and size in struct kernel_crypt_op

--- cryptodev_int.h |3 +++ cryptodev_main.c | 28 +++- 2 files changed, 14 insertions(+), 17 deletions(-) diff --git a/cryptodev_int.h b/cryptodev_int.h index 6b0b909..5072e8f 100644 --- a/cryptodev_int.h +++ b/cryptodev_int.h @@ -120,6 +120,9 @@ struct

[Cryptodev-linux-devel] [PATCH 8/8] add async versions of the code samples

100644 index 000..a288144 --- /dev/null +++ b/examples/async_speed.c @@ -0,0 +1,197 @@ +/* cryptodev_test - simple benchmark tool for cryptodev + * + *Copyright (C) 2010 by Phil Sutter phil.sut...@viprinet.com + * + * This program is free software; you can redistribute it and/or modify

[Cryptodev-linux-devel] [PATCH 6/8] increase the number of slots for asynchronous completion dynamically

--- cryptodev_main.c | 20 1 files changed, 16 insertions(+), 4 deletions(-) diff --git a/cryptodev_main.c b/cryptodev_main.c index 6ab235f..09d2c41 100644 --- a/cryptodev_main.c +++ b/cryptodev_main.c @@ -53,9 +53,10 @@ MODULE_LICENSE(GPL); #define CRYPTODEV_STATS

[Cryptodev-linux-devel] [PATCH 1/6] add support for CRYPTO_AES_ECB

--- cryptodev.h |1 + cryptodev_main.c |3 +++ 2 files changed, 4 insertions(+), 0 deletions(-) diff --git a/cryptodev.h b/cryptodev.h index adfc374..e16d30a 100644 --- a/cryptodev.h +++ b/cryptodev.h @@ -40,6 +40,7 @@ enum cryptodev_crypto_op_t { CRYPTO_SHA2_512_HMAC = 20,

[Cryptodev-linux-devel] [PATCH 2/4] crypto_run: cop-len is likely not zero

--- cryptodev_main.c |2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/cryptodev_main.c b/cryptodev_main.c index 38e1a44..69f2d29 100644 --- a/cryptodev_main.c +++ b/cryptodev_main.c @@ -730,7 +730,7 @@ static int crypto_run(struct fcrypt *fcr, struct kernel_crypt_op

[Cryptodev-linux-devel] [PATCH 3/4] minor whitespace cleanup

--- cryptodev_main.c |4 ++-- examples/hmac.c |4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/cryptodev_main.c b/cryptodev_main.c index 69f2d29..e869557 100644 --- a/cryptodev_main.c +++ b/cryptodev_main.c @@ -993,7 +993,7 @@ static int fill_kcop_from_cop(struct

[Cryptodev-linux-devel] [PATCH 1/4] since COP_FLAG_* are bits, we can OR them together for checks

--- cryptodev_main.c |3 +-- 1 files changed, 1 insertions(+), 2 deletions(-) diff --git a/cryptodev_main.c b/cryptodev_main.c index 2a6e4bc..38e1a44 100644 --- a/cryptodev_main.c +++ b/cryptodev_main.c @@ -705,8 +705,7 @@ static int crypto_run(struct fcrypt *fcr, struct kernel_crypt_op

[Cryptodev-linux-devel] [PATCH] speed, async_speed: fix measurement and units

Since measurement is done in bytes, use a capital B as unit. Also, stepping by 1024 from one magnitude to the next is more common. This is indicated by the 'i' in between the magnitude and unit symbols. --- examples/async_speed.c | 32 +++- examples/speed.c |

Re: [Cryptodev-linux-devel] mob broken?

Hi, On Fri, Dec 31, 2010 at 12:50:35AM +0200, Nikos Mavrogiannopoulos wrote: I've removed the mob branch few weeks before. I didn't find it much easy to use... Was it of use to you? Well, actually I didn't know about mob before getting in contact with cryptodev-linux. Although the idea of

[Cryptodev-linux-devel] [PATCH] allow updating the IV in userspace

When the user has specified COP_FLAG_WRITE_IV in crypt_op.flags, the updated IV will be written back to userspace. This is useful for encryption of continuous data in several steps, without having to care for each cipher's inerna. Protecting this functionality by a flag allows for backwards

[Cryptodev-linux-devel] [PATCH 1/5] enable CIOCGSESSINFO for compat-calls, too

--- cryptodev_main.c |1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/cryptodev_main.c b/cryptodev_main.c index cb81107..0f93010 100644 --- a/cryptodev_main.c +++ b/cryptodev_main.c @@ -1267,6 +1267,7 @@ cryptodev_compat_ioctl(struct file *file, unsigned int cmd, unsigned

[Cryptodev-linux-devel] [PATCH 2/5] alignment checks: prevent spurious warnings when ses-alignmask == 0

--- cryptodev_main.c |4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/cryptodev_main.c b/cryptodev_main.c index 0f93010..efd08c8 100644 --- a/cryptodev_main.c +++ b/cryptodev_main.c @@ -595,7 +595,7 @@ static int get_userbuf(struct csession *ses, struct kernel_crypt_op

[Cryptodev-linux-devel] [PATCH 1/2] tests: add cryptodev simulating openssl wrapper

Signed-off-by: Phil Sutter phil.sut...@viprinet.com --- tests/openssl_wrapper.c | 267 +++ 1 files changed, 267 insertions(+), 0 deletions(-) create mode 100644 tests/openssl_wrapper.c diff --git a/tests/openssl_wrapper.c b/tests/openssl_wrapper.c

[Cryptodev-linux-devel] [Patch v2 1/2] tests: add cryptodev simulating openssl wrapper

Signed-off-by: Phil Sutter phil.sut...@viprinet.com --- tests/openssl_wrapper.c | 267 +++ tests/openssl_wrapper.h |6 + 2 files changed, 273 insertions(+), 0 deletions(-) create mode 100644 tests/openssl_wrapper.c create mode 100644 tests

Re: [Cryptodev-linux-devel] (no subject)

Hi, On Wed, Feb 29, 2012 at 01:19:43PM +0100, Nikos Mavrogiannopoulos wrote: On Tue, Feb 28, 2012 at 11:56 PM, Phil Sutter p...@nwl.cc wrote: Another thing I just noticed, these commit-mails are somehow broken. E.g. backslashes are missing completely, and worse they're somehow interpreted

[Cryptodev-linux-devel] [PATCH 2/4] zc.c: simplify code a bit by exiting early

Also, hiding expressions in one-liners is not worth the lines saved. Signed-off-by: Phil Sutter phil.sut...@viprinet.com --- zc.c | 98 - 1 files changed, 48 insertions(+), 50 deletions(-) diff --git a/zc.c b/zc.c index 773dfb4

Re: [Cryptodev-linux-devel] Asynchronous support in Cryptodev is buggy

Hi, On Thu, Feb 14, 2013 at 05:20:07PM +, Dutta Yashpal-B05456 wrote: Looked at asynchronous support in Cryptodev-1.5 and found that it is getting in following flow cryptask_routine calling crypto_run. Now in crypto_run API, all the user-space process provided buffers are either being

Re: [Cryptodev-linux-devel] [PATCH] allow user override for kernel and installation directory

Hi, On Thu, Jun 27, 2013 at 05:39:59PM -0600, Anthony Foiani wrote: Cristian -- On Thu, Jun 27, 2013 at 7:42 AM, Cristian Stoica cristian.sto...@freescale.com wrote: - this is useful for cross-building for embedded systems This patch shouldn't be necessary: you can give variable

Re: [Cryptodev-linux-devel] [PATCH] Replace INIT_COMPLETION with reinit_completion.

Hi, On Wed, Jan 15, 2014 at 05:22:18PM +0200, Horia Geanta wrote: From: Cosmin Paraschiv cosmin.parasc...@freescale.com In the 3.12 Linux kernel, the INIT_COMPLETION macro has been replaced with an inline function, reinit_completion [1][2]. We are currently using the 3.13-rc3 Linux kernel,

Re: [Cryptodev-linux-devel] [PATCH] Replace INIT_COMPLETION with reinit_completion.

On Wed, Jan 15, 2014 at 06:21:30PM +0200, Horia Geantă wrote: On 1/15/2014 5:46 PM, Phil Sutter wrote: Hi, On Wed, Jan 15, 2014 at 05:22:18PM +0200, Horia Geanta wrote: From: Cosmin Paraschiv cosmin.parasc...@freescale.com In the 3.12 Linux kernel, the INIT_COMPLETION macro has been

Re: [Cryptodev-linux-devel] [PATCH] Replace INIT_COMPLETION with reinit_completion.

Hi Christian, On Tue, Jan 21, 2014 at 08:26:35AM +, cristian.sto...@freescale.com wrote: Hi Phil, Wouldn't it be better to conditionally define INIT_COMPLETION to reinit_completion for newer kernels? Doesn't this break downwards compatibility? I could think of something like:

Re: [Cryptodev-linux-devel] [PATCH v2] Replace INIT_COMPLETION with reinit_completion.

On Thu, Jan 16, 2014 at 12:18:08PM +0200, Horia Geanta wrote: From: Cosmin Paraschiv cosmin.parasc...@freescale.com In the 3.13-rc1 Linux kernel, the INIT_COMPLETION macro has been replaced with an inline function, reinit_completion [1][2]. We are currently using the 3.13-rc3 Linux kernel,

Re: [Cryptodev-linux-devel] [RFC 1/9] fix private data memleak

Hey Christian, On Mon, Jan 27, 2014 at 07:53:56AM +, cristian.sto...@freescale.com wrote: Why should that be necessary? The patch is concerned with 'close' being called with garbage input. I'm not sure about the call chain here: is 'close' never called when 'open' fails? What is the

Re: [Cryptodev-linux-devel] [PATCH] clean-up allocated resources after a failed open

Hi, Thanks for your patch, please see my comments below. On Tue, Feb 11, 2014 at 12:43:13PM +0200, Cristian Stoica wrote: If 'open /dev/crypto' fails, all allocated resources must be freed before open returns; close can't be called to clean-up since there is no file descriptor after a failed

Re: [Cryptodev-linux-devel] [PATCH v2] clean-up allocated resources after a failed open

Hi, On Wed, Feb 12, 2014 at 09:52:17AM +0200, Cristian Stoica wrote: If 'open /dev/crypto' fails, all allocated resources must be freed before open returns; close can't be called to clean-up since there is no file descriptor after a failed open. Signed-off-by: Cristian Stoica

Re: [Cryptodev-linux-devel] cryptodev uCLinux..

Brad, On Wed, Feb 12, 2014 at 06:20:31PM -0700, Brad Walker wrote: I am a Linux consultant working at a start-up over here in Boulder, CO. We are using uCLinux (based on Linux 2.6.39) on a really powerful processor (Freescale Kinetis) but it is resource limited. The chip has h/w

Re: [Cryptodev-linux-devel] [PATCH v2] clean-up allocated resources after a failed open

Hi Christian, On Thu, Feb 13, 2014 at 07:33:32AM +, cristian.sto...@freescale.com wrote: Hi Phil, for (i = 0; i DEF_COP_RINGSIZE; i++) { tmp = kzalloc(sizeof(struct todo_list_item), GFP_KERNEL); if (!tmp) - return -ENOMEM; +

Re: [Cryptodev-linux-devel] [PATCH v2] clean-up allocated resources after a failed open

Hi Christian, On Fri, Feb 14, 2014 at 09:00:07AM +, cristian.sto...@freescale.com wrote: Nope, that's exactly it. But in this case probably a mistake happened when submitting your v2, as it contains the above change. [] The v2 patch does not touch the pcr part anymore

Re: [Cryptodev-linux-devel] ? about cryptosession2

Hi, On Wed, Mar 05, 2014 at 10:01:16AM -0700, Brad Walker wrote: I noticed that on FreeBSD 7 OpenBSD cryptodev.h defines an ioctl called: CIOCGSESSION2 But, this is not defined in the cryptodev.h file for Linux. Curious to know why that is? Cryptodev-linux is meant to be compatible to

Re: [Cryptodev-linux-devel] [PATCH 1/3] use function-local storage for cipher and hmac keys

On Fri, May 30, 2014 at 01:59:02PM +0300, Cristian Stoica wrote: Composite ciphers (cipher + hmac) use both keys at the same time. This patch is the first in a series that adds support for composite ciphers keys. Signed-off-by: Cristian Stoica cristian.sto...@freescale.com All applied,

Re: [Cryptodev-linux-devel] [PATCH] fix tag and dst_len calculation for aead ciphers

On Thu, Jul 10, 2014 at 01:32:59PM +0300, Cristian Stoica wrote: Calculate tag and destination buffer length in a single place to avoid code duplication. The TLS case is fixed by rounding the destination length to cipher block size. Applied, thanks. Again, sorry for the delay. Cheers, Phil

Re: [Cryptodev-linux-devel] [PATCH] fix tag and dst_len calculation for aead ciphers

Hi Christian, On Mon, Aug 18, 2014 at 04:17:42PM +0300, Cristian Stoica wrote: Hi Phil, On 05.08.2014 01:36, Phil Sutter wrote: On Thu, Jul 10, 2014 at 01:32:59PM +0300, Cristian Stoica wrote: Calculate tag and destination buffer length in a single place to avoid code duplication

Re: [Cryptodev-linux-devel] cryptodev-linux 1.7

I *knew* it. On Sat, Feb 07, 2015 at 11:01:05PM +0100, Phil Sutter wrote: I have eventually managed to release a new version of cryptodev-linux. As almost a year has passed since the last release, the list of changes is a rather long one: There is the bug already: It's 2015, so almost two

[Cryptodev-linux-devel] cryptodev-linux 1.7

Hi, I have eventually managed to release a new version of cryptodev-linux. As almost a year has passed since the last release, the list of changes is a rather long one: * Added support for composite AEAD keys by Cristian Stoica. * Added support for sysctl to modify verbosity by Nikolaos

Re: [Cryptodev-linux-devel] Python bindings for the cryptodev module.

Hi, On Wed, Mar 04, 2015 at 05:50:35PM +0200, Tilemachos Charalampous wrote: Hello. I'm an undergraduate student studying electrical and computer engineering at the NTUA. As part of an Operating Systems course, I worked a bit on the cryptodev module, and needed it to use it in a Pyhton

Re: [Cryptodev-linux-devel] cryptodev, mv_cesa sha1

Hello Jan, On Mon, Apr 27, 2015 at 12:44:50PM +0200, JM wrote: I installed cryptodev 1.7 on my Debian box running on Marvell Feroceon processor (featuring mv_cesa hardware accelerator). I'm running debian wheezy with openssl_1.0.1e (with enabled cryptodev and the two recommended patches) and

Re: [Cryptodev-linux-devel] Issue with make check

Hi, On Fri, Apr 10, 2015 at 11:08:10AM +0200, Filip Kubicz wrote: The version of cc is cc (Ubuntu 4.8.2-18ubuntu1) 4.8.2 Copyright (C) 2013 Free Software Foundation, Inc. OK, looks like GCC as well. According to the GCC docs at https://gcc.gnu.org/onlinedocs/gcc/Link-Options.html

Re: [Cryptodev-linux-devel] Problem with OpenSSH/OpenSSL Interaction When Cryptodev is Used

On Wed, May 27, 2015 at 11:20:28AM +0100, Gordan Bobic wrote: On 2015-05-27 11:07, Phil Sutter wrote: Hi, On Wed, May 27, 2015 at 10:25:00AM +0100, Gordan Bobic wrote: ./cipher-aead-srtp ioctl(CIOCGSESSION): Invalid argument Not sure if this should succeed at all, at least that's

Re: [Cryptodev-linux-devel] cryptodev fails to compile with kernel 4.04 on arch

Hi, On Mon, Jun 01, 2015 at 09:55:16PM +0200, Sebastian Anding wrote: i tried to compile cryptodev 1.7 on arch linux at a beagle bone black. I attached a uname -a, gcc --version and make 2 compile.err and 1 compile.err.add This looks like a problem in your build system, maybe you are

Re: [Cryptodev-linux-devel] [PATCH] remove code duplication in cryptodev_hash_init

On Tue, Jan 12, 2016 at 10:16:18AM +0200, Cristian Stoica wrote: > cryptodev_hash_init is concerned mostly with allocating data structures > for hash operations. > This patch replaces the call it makes to crypto_ahash_init with > one to cryptodev_hash_reset to avoid code duplication. This call is

Re: [Cryptodev-linux-devel] [PATCH] fix the NULL input or output case for get_userbuf

Hi, On Tue, Jan 12, 2016 at 12:08:02PM +0200, Cristian Stoica wrote: > NULL input or output is a valid case and we should not attempt > to write default values to NULL input or output pointers. > > Signed-off-by: Cristian Stoica > --- > zc.c | 3 --- > 1 file changed,

Re: [Cryptodev-linux-devel] [PATCH] fix the NULL input or output case for get_userbuf

On Tue, Jan 19, 2016 at 01:38:53PM +, Cristian Stoica wrote: > > > On 01/18/2016 11:04 PM, Phil Sutter wrote: > > On Tue, Jan 12, 2016 at 12:08:02PM +0200, Cristian Stoica wrote: > ... > >> - *src_sg = NULL; /* default to no input */ > >> - *dst_sg

Re: [Cryptodev-linux-devel] [PATCH] fix the NULL input or output case for get_userbuf

Hi Cristian, On Wed, Jan 20, 2016 at 02:06:49PM +, Cristian Stoica wrote: > > I really have no idea where you're heading at with this. But without further > > context, it doesn't make any sense to me. If you have a functional > > enhancement up your sleeve, I suggest submitting that along

Re: [Cryptodev-linux-devel] Crypto API changes in 4.8

On Sat, Aug 06, 2016 at 01:22:01AM +0200, Michael Weiser wrote: > Hi Phil, > > On Fri, Aug 05, 2016 at 10:09:08PM +0200, Phil Sutter wrote: > > > > I've done some groundwork, mostly research through a well known search > > > engine and in the kernel code. What I'

Re: [Cryptodev-linux-devel] Question regarding cryptodev using hardware acceleration for FIPS validation

Hi Chirag, On Tue, Jan 31, 2017 at 10:58:15AM -0800, Chirag Shahani wrote: [...] > However, when I running the same user space program on a box which has SoC > C2000 co-processor (QAT-1.5), I was expecting the code to* NOT *print > printf("Note: This is not an accelerated cipher\n"). > > > > *

Re: [Cryptodev-linux-devel] Question regarding cryptodev using hardware acceleration for FIPS validation

Hi, On Thu, Feb 02, 2017 at 06:53:55AM -0800, Chirag Shahani wrote: > Thanks for your reply. > > Yes. The QAT 1.5 module is loaded. I verified that by: > > cat /proc/icp_c2xxx_dev0/version > +--+ > | Hardware and Software versions for device 0 > |

Re: [Cryptodev-linux-devel] [PATCH] adjust to API changes in kernel >=4.10

Hi, On Thu, Feb 09, 2017 at 08:28:06AM +, Cristian Stoica wrote: > I'm sure I can find an explanation of why I've written it that way but as I > look at your version I wonder why I didn't recheck my assumptions and > simplify it further. > > Anyway, Phil can apply your suggested version

Re: [Cryptodev-linux-devel] [PATCH] Fix ablkcipher algorithms usage in v4.8+ kernels

Hi, On Wed, Feb 08, 2017 at 04:41:50PM +, Horia Geantă wrote: > Phil? I intentionally left this out since (according to my INBOX) there are open questions from Michael. Did you sort this out in private? Cheers, Phil ___ Cryptodev-linux-devel

Re: [Cryptodev-linux-devel] [PATCH] Fix ablkcipher algorithms usage in v4.8+ kernels

On Thu, Feb 09, 2017 at 11:47:16AM +, Horia Geantă wrote: > On 2/9/2017 12:35 PM, Phil Sutter wrote: > > Hi, > > > > On Wed, Feb 08, 2017 at 04:41:50PM +, Horia Geantă wrote: > >> Phil? > > > > I intentionally left this out since (according to m

Re: [Cryptodev-linux-devel] [PATCH 2/6] fix warnings of "implicit declaration of function" in async_speed

On Wed, Oct 26, 2016 at 02:09:25PM +0300, Cristian Stoica wrote: > Signed-off-by: Cristian Stoica > --- > tests/async_speed.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/tests/async_speed.c b/tests/async_speed.c > index 1188599..dad5bc5 100644 > ---

Re: [Cryptodev-linux-devel] [PATCH 6/6] setting KERNEL_DIR is not necessary to build tests

On Wed, Oct 26, 2016 at 02:09:29PM +0300, Cristian Stoica wrote: > Signed-off-by: Cristian Stoica > --- > tests/Makefile | 2 -- > 1 file changed, 2 deletions(-) > > diff --git a/tests/Makefile b/tests/Makefile > index 49959de..2502f32 100644 > --- a/tests/Makefile >

Re: [Cryptodev-linux-devel] [PATCH 3/6] remove not used local variables

On Wed, Oct 26, 2016 at 02:09:26PM +0300, Cristian Stoica wrote: > Signed-off-by: Cristian Stoica > --- > tests/async_hmac.c | 2 -- > tests/cipher_comp.c | 2 +- > tests/hash_comp.c | 5 ++--- > tests/hmac.c| 2 -- > tests/sha_speed.c | 1 - > tests/speed.c

Re: [Cryptodev-linux-devel] [PATCH 1/6] avoid implicit conversion between signed and unsigned char

Hi, On Wed, Oct 26, 2016 at 02:09:24PM +0300, Cristian Stoica wrote: > Signed-off-by: Cristian Stoica > --- > tests/async_cipher.c | 36 ++-- > tests/async_hmac.c | 8 +++ > tests/cipher-aead-srtp.c | 50

[Cryptodev-linux-devel] tests/cipher-gcm crashes kernel

Hi, Testing cryptodev on a recent kernel (4.10.0), I get an oops when running tests/cipher-gcm. It fails with the last entry in aes_gcm_vectors (the one with auth data). I couldn't find out yet where the problem comes from, but looking at cryptodev_cipher_auth(), it seems like we miss attaching

[Cryptodev-linux-devel] cryptodev-linux 1.9

Hi, I just released current master as new version 1.9. It is comprised of the following commits in addition to the previous release: * fix benchmarks linking * fix Makefile to allow parallel make with -j option * use Linux kernel conventions for Makefile variables * for consistency, use $(...)