Re: [PATCH] af_alg: disable kernel hash functions by default

[Bruno Haible]

Pádraig Brady wrote:
> > All the kernel routines were seen to be significantly slower
> > with these relatively recent components on an i3-2310M system:
> >   kernel-4.10.6-200.fc25.x86_64
> >   openssl-1.0.2m-1.fc25.x86_64
> > sha1 was nearly twice as slow in the kernel for example.
> 
> Assaf Gordon has just detailed more reasons why af_alg
> should not be enabled by default at: [7]

While I think that the performance arguments are not good arguments
until a method to disable the slow code paths (namely, read /proc/crypto
at run time) [1] is implemented,

and I did not find Assaf's earlier arguments [2][3][4][5][6] convincing,

and point 1 from [7] is merely a bug that ought to be fixed,

*** the points 2 and 3 from [7] are very valid points. ***

Please go ahead. The proposed patch looks fine.

Bruno

[1] https://lists.gnu.org/archive/html/bug-gnulib/2018-05/msg00049.html
[2] https://lists.gnu.org/archive/html/bug-gnulib/2018-05/msg00041.html
[3] https://lists.gnu.org/archive/html/bug-gnulib/2018-05/msg00052.html
[4] https://lists.gnu.org/archive/html/bug-gnulib/2018-05/msg00053.html
[5] https://lists.gnu.org/archive/html/bug-gnulib/2018-05/msg00054.html
[6] https://lists.gnu.org/archive/html/bug-gnulib/2018-05/msg00063.html
[7] https://lists.gnu.org/archive/html/coreutils/2018-06/msg00034.html