Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
On Mon, Jan 03, 2011 at 08:29:02PM +, Ben Hutchings wrote: On Mon, 2011-01-03 at 10:12 -0700, dann frazier wrote: On Mon, Jan 03, 2011 at 05:05:32AM +, Ben Hutchings wrote: On Sun, 2011-01-02 at 21:30 -0700, dann frazier wrote: On Sun, Jan 02, 2011 at 08:57:19PM +, Ben Hutchings wrote: [...] Please can you test whether aesni-intel loads and works in: 1. Package version 2.6.32-12 http://snapshot.debian.org/package/linux-2.6/2.6.32-12/. This was the last version with aesni-intel included. 2. Package version 2.6.32-29, modified to reenable aesni-intel. See the instructions at http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official. a. Follow section 4.2.1. b. Change '# CONFIG_CRYPTO_AES_NI_INTEL is not set' to 'CONFIG_CRYPTO_AES_NI_INTEL=m' in debian/config/kernelarch-x86/config-arch-64. c. Follow section 4.2.4. fyi, 2.6.32-12 2.6.32-29 w/ CONFIG_CRYPTO_AES_NI_INTEL=m both boot successfully w/ the aesni_intel module on my Lenovo T410. I'm guessing you don't have an AES-encrypted hard drive though... I do: $ sudo cryptsetup status sda5_crypt /dev/mapper/sda5_crypt is active: cipher: aes-cbc-essiv:sha256 [...] Though I don't know how to verify that the hardware implementation is actually being used. I think that trying to remove the module is a valid test. Finally found a convenient time to reboot; in both cases (-12 -29 + the module) the module is unremovable on my system. -dann -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110110203956.ga2...@dannf.org
Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
On Mon, Jan 03, 2011 at 05:05:32AM +, Ben Hutchings wrote: On Sun, 2011-01-02 at 21:30 -0700, dann frazier wrote: On Sun, Jan 02, 2011 at 08:57:19PM +, Ben Hutchings wrote: [...] Please can you test whether aesni-intel loads and works in: 1. Package version 2.6.32-12 http://snapshot.debian.org/package/linux-2.6/2.6.32-12/. This was the last version with aesni-intel included. 2. Package version 2.6.32-29, modified to reenable aesni-intel. See the instructions at http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official. a. Follow section 4.2.1. b. Change '# CONFIG_CRYPTO_AES_NI_INTEL is not set' to 'CONFIG_CRYPTO_AES_NI_INTEL=m' in debian/config/kernelarch-x86/config-arch-64. c. Follow section 4.2.4. fyi, 2.6.32-12 2.6.32-29 w/ CONFIG_CRYPTO_AES_NI_INTEL=m both boot successfully w/ the aesni_intel module on my Lenovo T410. I'm guessing you don't have an AES-encrypted hard drive though... I do: $ sudo cryptsetup status sda5_crypt /dev/mapper/sda5_crypt is active: cipher: aes-cbc-essiv:sha256 [...] Though I don't know how to verify that the hardware implementation is actually being used. fwiw, the init messages do differ for me between -12 -29 (as the git log suggests it should): -12: [ 13.565523] alg: No test for __aes-aesni (__driver-aes-aesni) [ 13.565542] alg: No test for __ecb-aes-aesni (__driver-ecb-aes-aesni) [ 13.565559] alg: No test for __cbc-aes-aesni (__driver-cbc-aes-aesni) [ 13.566877] alg: No test for __ecb-aes-aesni (cryptd(__driver-ecb-aes-aesni)) [ 13.569514] alg: No test for __cbc-aes-aesni (cryptd(__driver-cbc-aes-aesni)) -29: [ 10.766143] alg: No test for __cbc-aes-aesni (cryptd(__driver-cbc-aes-aesni)) -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110103171252.gb12...@dannf.org
Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
What has to be done until the proposed solution to blacklist the module could be implemented? I do think that it is not very nice to have a big part of those with a newer amd64 box unable to use this functionality. If I get it correctly, the main reason for the decision to disable that module is that it's not playing nice with a T400. What is the number of the bug report with upstream on that matter? Best Regards, Christian Franke -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110102194127.gb19...@ts2server.com
Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
On Sun, 2011-01-02 at 20:41 +0100, cfchris6 wrote: What has to be done until the proposed solution to blacklist the module could be implemented? I do think that it is not very nice to have a big part of those with a newer amd64 box unable to use this functionality. If I get it correctly, the main reason for the decision to disable that module is that it's not playing nice with a T400. What is the number of the bug report with upstream on that matter? I don't know of one. There is a Fedora bug report https://bugzilla.redhat.com/show_bug.cgi?id=571577 and if I read that correctly then this is not a model-specific bug - the module was actually completely broken in 2.6.32 if CONFIG_CRYPTO_FIPS=y (which it is in Debian packages). However, it also looks like this was fixed in 2.6.32.19, which is incorporated in the source for the current package. Please can you test whether aesni-intel loads and works in: 1. Package version 2.6.32-12 http://snapshot.debian.org/package/linux-2.6/2.6.32-12/. This was the last version with aesni-intel included. 2. Package version 2.6.32-29, modified to reenable aesni-intel. See the instructions at http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official. a. Follow section 4.2.1. b. Change '# CONFIG_CRYPTO_AES_NI_INTEL is not set' to 'CONFIG_CRYPTO_AES_NI_INTEL=m' in debian/config/kernelarch-x86/config-arch-64. c. Follow section 4.2.4. Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse. signature.asc Description: This is a digitally signed message part
Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
On Sun, Jan 02, 2011 at 08:57:19PM +, Ben Hutchings wrote: On Sun, 2011-01-02 at 20:41 +0100, cfchris6 wrote: What has to be done until the proposed solution to blacklist the module could be implemented? I do think that it is not very nice to have a big part of those with a newer amd64 box unable to use this functionality. If I get it correctly, the main reason for the decision to disable that module is that it's not playing nice with a T400. What is the number of the bug report with upstream on that matter? I don't know of one. There is a Fedora bug report https://bugzilla.redhat.com/show_bug.cgi?id=571577 and if I read that correctly then this is not a model-specific bug - the module was actually completely broken in 2.6.32 if CONFIG_CRYPTO_FIPS=y (which it is in Debian packages). However, it also looks like this was fixed in 2.6.32.19, which is incorporated in the source for the current package. Please can you test whether aesni-intel loads and works in: 1. Package version 2.6.32-12 http://snapshot.debian.org/package/linux-2.6/2.6.32-12/. This was the last version with aesni-intel included. 2. Package version 2.6.32-29, modified to reenable aesni-intel. See the instructions at http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official. a. Follow section 4.2.1. b. Change '# CONFIG_CRYPTO_AES_NI_INTEL is not set' to 'CONFIG_CRYPTO_AES_NI_INTEL=m' in debian/config/kernelarch-x86/config-arch-64. c. Follow section 4.2.4. fyi, 2.6.32-12 2.6.32-29 w/ CONFIG_CRYPTO_AES_NI_INTEL=m both boot successfully w/ the aesni_intel module on my Lenovo T410. -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110103043033.ga2...@dannf.org
Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
On Sun, 2011-01-02 at 21:30 -0700, dann frazier wrote: On Sun, Jan 02, 2011 at 08:57:19PM +, Ben Hutchings wrote: [...] Please can you test whether aesni-intel loads and works in: 1. Package version 2.6.32-12 http://snapshot.debian.org/package/linux-2.6/2.6.32-12/. This was the last version with aesni-intel included. 2. Package version 2.6.32-29, modified to reenable aesni-intel. See the instructions at http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official. a. Follow section 4.2.1. b. Change '# CONFIG_CRYPTO_AES_NI_INTEL is not set' to 'CONFIG_CRYPTO_AES_NI_INTEL=m' in debian/config/kernelarch-x86/config-arch-64. c. Follow section 4.2.4. fyi, 2.6.32-12 2.6.32-29 w/ CONFIG_CRYPTO_AES_NI_INTEL=m both boot successfully w/ the aesni_intel module on my Lenovo T410. I'm guessing you don't have an AES-encrypted hard drive though... Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse. signature.asc Description: This is a digitally signed message part
Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
Package: linux-headers-2.6.32-5-amd64 Version: 2.6.32-23 Severity: normal Since 2.6.32-5 the aesni-intel modul is missing in the kernel, therefore crypto-stuff on intel i5/i7 with aes-ni is a lot slower than it should be. Would be nice if it could be activated per default again. Cryptgraphic API - AES cipher algorithms (AES-NI) lsmod |grep -i aes aes_x86_64 7340 3 aes_generic25714 1 aes_x86_64 There should be the aesni-intel-module -- System Information: Debian Release: squeeze/sid APT prefers testing APT policy: (500, 'testing') Architecture: amd64 (x86_64) Kernel: Linux 2.6.32-5-amd64 (SMP w/4 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=C (charmap=UTF-8) (ignored: LC_ALL set to en_US.UTF-8) Shell: /bin/sh linked to /bin/dash Versions of packages linux-headers-2.6.32-5-amd64 depends on: ii gcc-4.3 4.3.5-2The GNU C compiler ii linux-headers-2.6.32-5-common 2.6.32-23 Common header files for Linux 2.6. ii linux-kbuild-2.6.32 2.6.32-1 Kbuild infrastructure for Linux 2. linux-headers-2.6.32-5-amd64 recommends no packages. linux-headers-2.6.32-5-amd64 suggests no packages. -- no debconf information -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100921212402.23621.15564.report...@blubba
Bug#597658: linux-headers-2.6.32-5-amd64: Missing aesni-intel module in kernel
On Tue, 2010-09-21 at 23:24 +0200, Benjamin Steininger wrote: Package: linux-headers-2.6.32-5-amd64 Version: 2.6.32-23 Severity: normal Since 2.6.32-5 the aesni-intel modul is missing in the kernel, therefore crypto-stuff on intel i5/i7 with aes-ni is a lot slower than it should be. Would be nice if it could be activated per default again. Cryptgraphic API - AES cipher algorithms (AES-NI) lsmod |grep -i aes aes_x86_64 7340 3 aes_generic25714 1 aes_x86_64 There should be the aesni-intel-module [...] From the changelog: * x86: Disable CRYPTO_AES_NI_INTEL as it causes boot failures on T410. Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse. signature.asc Description: This is a digitally signed message part