Unit of /proc/sys/kernel/random/poolsize

Unit of /proc/sys/kernel/random/poolsize

[Tom Gundersen]

Hi,

This might not be so much a bug report, as a request for clarification.

In random(4) it is explained that /proc/sys/kernel/random/poolsize
contains the value in bits or bytes in >=2.6 and 2.4 respectively.
However, the example in the preceding paragraph assumes a 2.4 kernel
is being used, and treats the value as if it were in bytes.

I ran across this as I maintain the initscripts for Arch Linux, where
we have been (mistakenly) following the example even though we use a
3.0 kernel. Maybe it would be worthwhile to change the example into
the 2.6 interface, or at least include a word of caution?

I realise that using a value which is too large, probably does no harm
at all, but not being cryptography expert, I'd rather not take the
chance.

Cheers,

Tom
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Unit of /proc/sys/kernel/random/poolsize

[Michael Kerrisk (man-pages)]

Hello Tom,

On 10/22/2011 02:13 PM, Tom Gundersen wrote:
> Hi,
> 
> This might not be so much a bug report, as a request for clarification.
> 
> In random(4) it is explained that /proc/sys/kernel/random/poolsize
> contains the value in bits or bytes in >=2.6 and 2.4 respectively.
> However, the example in the preceding paragraph assumes a 2.4 kernel
> is being used, and treats the value as if it were in bytes.
> 
> I ran across this as I maintain the initscripts for Arch Linux, where
> we have been (mistakenly) following the example even though we use a
> 3.0 kernel. Maybe it would be worthwhile to change the example into
> the 2.6 interface, or at least include a word of caution?
> 
> I realise that using a value which is too large, probably does no harm
> at all, but not being cryptography expert, I'd rather not take the
> chance.

Long after the fact... I applied the patch below.
Okay?

Thanks for the report.

Cheers,

Michael

PS I am going to be in Oslo a couple of times in coming months.
Perhaps we could meet up for a beer or so.

diff --git a/man4/random.4 b/man4/random.4
index 2519981..4a32aac 100644
--- a/man4/random.4
+++ b/man4/random.4
@@ -171,7 +171,7 @@ This reduces the actual amount of noise in the entropy pool
 below the estimate.
 In order to counteract this effect, it helps to carry
 entropy pool information across shut-downs and start-ups.
-To do this, add the following lines to an appropriate script
+To do this, add the lines to an appropriate script
 which is run during the Linux system start-up sequence:
 
 .nf
@@ -186,7 +186,8 @@ which is run during the Linux system start-up sequence:
     fi
     chmod 600 $random_seed
     poolfile=/proc/sys/kernel/random/poolsize
-    [ \-r $poolfile ] && bytes=\`cat $poolfile\` || bytes=512
+    [ \-r $poolfile ] && bits=\`cat $poolfile\` || bits=4096
+    bytes=$(expr $bits / 8)
     dd if=/dev/urandom of=$random_seed count=1 bs=$bytes
 .fi
 
@@ -201,9 +202,14 @@ run during the Linux system shutdown:
     touch $random_seed
     chmod 600 $random_seed
     poolfile=/proc/sys/kernel/random/poolsize
-    [ \-r $poolfile ] && bytes=\`cat $poolfile\` || bytes=512
+    [ \-r $poolfile ] && bits=\`cat $poolfile\` || bits=4096
+    bytes=$(expr $bits / 8)
     dd if=/dev/urandom of=$random_seed count=1 bs=$bytes
 .fi
+
+In the above examples, we assume Linux 2.6.0 or later, where
+.IR /proc/sys/kernel/random/poolsize
+returns the size of the entropy pool in bits (see below).
 .SS /proc Interface
 The files in the directory
 .I /proc/sys/kernel/random


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Unit of /proc/sys/kernel/random/poolsize

[walter harms]

Am 17.12.2015 08:51, schrieb Michael Kerrisk (man-pages):
> Hello Tom,
> 
> On 10/22/2011 02:13 PM, Tom Gundersen wrote:
>> Hi,
>>
>> This might not be so much a bug report, as a request for clarification.
>>
>> In random(4) it is explained that /proc/sys/kernel/random/poolsize
>> contains the value in bits or bytes in >=2.6 and 2.4 respectively.
>> However, the example in the preceding paragraph assumes a 2.4 kernel
>> is being used, and treats the value as if it were in bytes.
>>
>> I ran across this as I maintain the initscripts for Arch Linux, where
>> we have been (mistakenly) following the example even though we use a
>> 3.0 kernel. Maybe it would be worthwhile to change the example into
>> the 2.6 interface, or at least include a word of caution?
>>
>> I realise that using a value which is too large, probably does no harm
>> at all, but not being cryptography expert, I'd rather not take the
>> chance.
> 
> Long after the fact... I applied the patch below.
> Okay?
> 
> Thanks for the report.
> 
> Cheers,
> 
> Michael
> 
> PS I am going to be in Oslo a couple of times in coming months.
> Perhaps we could meet up for a beer or so.
> 
> diff --git a/man4/random.4 b/man4/random.4
> index 2519981..4a32aac 100644
> --- a/man4/random.4
> +++ b/man4/random.4
> @@ -171,7 +171,7 @@ This reduces the actual amount of noise in the entropy pool
>  below the estimate.
>  In order to counteract this effect, it helps to carry
>  entropy pool information across shut-downs and start-ups.
> -To do this, add the following lines to an appropriate script
> +To do this, add the lines to an appropriate script
>  which is run during the Linux system start-up sequence:
>  
>  .nf
> @@ -186,7 +186,8 @@ which is run during the Linux system start-up sequence:
>      fi
>      chmod 600 $random_seed
>      poolfile=/proc/sys/kernel/random/poolsize
> -    [ \-r $poolfile ] && bytes=\`cat $poolfile\` || bytes=512
> +    [ \-r $poolfile ] && bits=\`cat $poolfile\` || bits=4096

I believe it was end of the 80's when $( ) was introduced to replace \`
Since this example is for 2.6 maybe it would be useful to use it and
improve readability ?

re,
 wh


> +    bytes=$(expr $bits / 8)
>      dd if=/dev/urandom of=$random_seed count=1 bs=$bytes
>  .fi
>  
> @@ -201,9 +202,14 @@ run during the Linux system shutdown:
>      touch $random_seed
>      chmod 600 $random_seed
>      poolfile=/proc/sys/kernel/random/poolsize
> -    [ \-r $poolfile ] && bytes=\`cat $poolfile\` || bytes=512
> +    [ \-r $poolfile ] && bits=\`cat $poolfile\` || bits=4096
> +    bytes=$(expr $bits / 8)
>      dd if=/dev/urandom of=$random_seed count=1 bs=$bytes
>  .fi
> +
> +In the above examples, we assume Linux 2.6.0 or later, where
> +.IR /proc/sys/kernel/random/poolsize
> +returns the size of the entropy pool in bits (see below).
>  .SS /proc Interface
>  The files in the directory
>  .I /proc/sys/kernel/random
> 
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Unit of /proc/sys/kernel/random/poolsize

[Michael Kerrisk (man-pages)]

Hello Walter,

On 12/17/2015 09:56 PM, walter harms wrote:
> 
> 
> Am 17.12.2015 08:51, schrieb Michael Kerrisk (man-pages):
>> Hello Tom,
>>
>> On 10/22/2011 02:13 PM, Tom Gundersen wrote:
>>> Hi,
>>>
>>> This might not be so much a bug report, as a request for clarification.
>>>
>>> In random(4) it is explained that /proc/sys/kernel/random/poolsize
>>> contains the value in bits or bytes in >=2.6 and 2.4 respectively.
>>> However, the example in the preceding paragraph assumes a 2.4 kernel
>>> is being used, and treats the value as if it were in bytes.
>>>
>>> I ran across this as I maintain the initscripts for Arch Linux, where
>>> we have been (mistakenly) following the example even though we use a
>>> 3.0 kernel. Maybe it would be worthwhile to change the example into
>>> the 2.6 interface, or at least include a word of caution?
>>>
>>> I realise that using a value which is too large, probably does no harm
>>> at all, but not being cryptography expert, I'd rather not take the
>>> chance.
>>
>> Long after the fact... I applied the patch below.
>> Okay?
>>
>> Thanks for the report.
>>
>> Cheers,
>>
>> Michael
>>
>> PS I am going to be in Oslo a couple of times in coming months.
>> Perhaps we could meet up for a beer or so.
>>
>> diff --git a/man4/random.4 b/man4/random.4
>> index 2519981..4a32aac 100644
>> --- a/man4/random.4
>> +++ b/man4/random.4
>> @@ -171,7 +171,7 @@ This reduces the actual amount of noise in the entropy pool
>>  below the estimate.
>>  In order to counteract this effect, it helps to carry
>>  entropy pool information across shut-downs and start-ups.
>> -To do this, add the following lines to an appropriate script
>> +To do this, add the lines to an appropriate script
>>  which is run during the Linux system start-up sequence:
>>  
>>  .nf
>> @@ -186,7 +186,8 @@ which is run during the Linux system start-up sequence:
>>      fi
>>      chmod 600 $random_seed
>>      poolfile=/proc/sys/kernel/random/poolsize
>> -    [ \-r $poolfile ] && bytes=\`cat $poolfile\` || bytes=512
>> +    [ \-r $poolfile ] && bits=\`cat $poolfile\` || bits=4096
> 
> I believe it was end of the 80's when $( ) was introduced to replace \`
> Since this example is for 2.6 maybe it would be useful to use it and
> improve readability ?

Yes, of course, and especially so since I use the modern syntax in the 
new piece that I added. Fixed. Thanks for catching the obvious point
that I missed!

Cheers,

Michael



-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html