Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 20 Jun 2006 23:24:24 +0300
From:      Giorgos Keramidas <keramida@FreeBSD.org>
To:        Jason Evans <jasone@FreeBSD.org>, Dan Lukes <dan@obluda.cz>
Cc:        freebsd-security@FreeBSD.org
Subject:   Re: memory pages nulling when releasing
Message-ID:  <20060620202424.GB11347@gothmog.pc>
In-Reply-To: <44985028.6020901@FreeBSD.org> <44984A25.8060001@obluda.cz> <449833DD.70904@FreeBSD.org>
References:  <44980ECC.5000607@obluda.cz> <20060620170353.GB90842@gothmog.pc> <449833DD.70904@FreeBSD.org> <44984A25.8060001@obluda.cz> <20060618203903.31161.qmail@web30306.mail.mud.yahoo.com> <44967861.6070509@obluda.cz> <20060620125350.10d0c9ef.nick@nickwithers.com> <44980ECC.5000607@obluda.cz> <20060620170353.GB90842@gothmog.pc> <449833DD.70904@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2006-06-20 10:43, Jason Evans <jasone@FreeBSD.org> wrote:
>Giorgos Keramidas wrote:
>>On 2006-06-20 17:05, Dan Lukes <dan@obluda.cz> wrote:
>>>BTW, Z nor J doesn't imply 'R' despite of text of manual page.
>>
>> Are you sure?  Then we should fix the manpage.
>> Cc'ing Jason Evans, who may be interested too.
> 
> The man page looks accurate to me.  What aspect of the descriptions of 
> 'J' and 'Z' do you think is incorrect?

On 2006-06-20 21:19, Dan Lukes <dan@obluda.cz> wrote:
> 	From manual page:
>  =======================
> J       ...
> 	This options also sets the ``R'' option.
> 	...
> Z       This option implicitly sets the ``J'' and ``R'' options,
>  =======================
> 
> 
> 	From libc/stdlib/malloc.c (the lines containing malloc_realloc + 
> 	some lines around)
>  =======================
> static int malloc_realloc;
> ...
> 		case 'r': malloc_realloc = 0; break;
> 		case 'R': malloc_realloc = 1; break;
> 		case 'j': malloc_junk    = 0; break;
> ...
>         if (!malloc_realloc && 			/* Unless we have 
>         to, */
> 	  size <= osize && 			/* .. or are too small, ...
> 	if (!malloc_realloc &&		/* Unless we have to, */
> 	  size <= osize && 		/* ..or are too small, */
>  =======================
> 
> The malloc_realloc seems to be set on one place only - when R/r
> present. Then it is used only. I see no dependency to J nor Z
> 
> I send no PR because I have submitted several more important PR's
> which are untouched for months. I don't want to overload committers
> nor PR database.

No, that's ok.  It's better to send what you have found so it won't be
missed forever :-)

On 2006-06-20 12:44, Jason Evans <jasone@freebsd.org> wrote:
> Ah, this is for RELENG_6.  I had assumed you were talking about
> -current.  I agree with you that the man page is inaccurate for
> RELENG_6.  I don't currently use RELENG_6 for development though, so
> I'd recommend submitting a PR with a patch assigned to the docs folks
> if you want it fixed.  In my experience they are very responsive where
> man page accuracy is concerned.

Sure!

Dan, if you submit a PR, then feel free to include my own address at the
Cc: line of send-pr's report.  Then I'll grab the PR and work through
the necessary changes for RELENG_6 with you.

Thanks for bringing this manpage bug to our attention :)

- Giorgos




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060620202424.GB11347>