On Mon, Jun 10, 2019 at 09:00:34AM -0400, Shawn Webb wrote:
> On Mon, Jun 10, 2019 at 03:07:11AM +, Doug Moore wrote:
> > Author: dougm
> > Date: Mon Jun 10 03:07:10 2019
> > New Revision: 348843
> > URL: https://svnweb.freebsd.org/changeset/base/348843
> >
> > Log:
> > There are times when
On Tue, Jun 11, 2019 at 01:33:23AM +1000, Bruce Evans wrote:
> On Mon, 10 Jun 2019, Shawn Webb wrote:
>
> > On Mon, Jun 10, 2019 at 03:07:11AM +, Doug Moore wrote:
> > > ...
> > > Log:
> > > There are times when a len==0 parameter to mmap is okay. But on a
> > > 32-bit machine, a len param
On Mon, 10 Jun 2019, Shawn Webb wrote:
On Mon, Jun 10, 2019 at 03:07:11AM +, Doug Moore wrote:
...
Log:
There are times when a len==0 parameter to mmap is okay. But on a
32-bit machine, a len parameter just a few bytes short of 4G, rounded
up to a page boundary and hitting zero then,
The comment and the code that rejects size==0, or doesn't, are copied
below. Konstantin Belousov is the last person to have touched most of
it, and can better explain its meaning than I.
Doug Moore
/*
* Enforce the constraints.
* Mapping of length 0 is only allowed for old binaries
Sounds good! I think the manpage still might still need a change
to match the current behavior, or perhaps matching something similar
to that vm_mmap.c comment. But that comment brings another question:
what's the definition of "old binaries"? a.out?
Thanks,
--
Shawn Webb
Cofounder / Security En
This comment appears in vm_mmap.c:
* Mapping of length 0 is only allowed for old binaries.
and my intent was to say, to whoever wrote that comment, that I was not
disallowing the mapping of length zero with this change. I was only
intending to affect a case in which the length was transform
On Mon, Jun 10, 2019 at 03:07:11AM +, Doug Moore wrote:
> Author: dougm
> Date: Mon Jun 10 03:07:10 2019
> New Revision: 348843
> URL: https://svnweb.freebsd.org/changeset/base/348843
>
> Log:
> There are times when a len==0 parameter to mmap is okay. But on a
> 32-bit machine, a len param
On 10/06/2019 09:42, Doug Moore wrote:
> -fwrapv concerns signed arithmetic. This calculation is with unsigned
> arithmetic, and the possibility of wrapping around to 0 is part of the
> language.
Oh, sorry for the noise!
> On 6/10/19 1:35 AM, Andriy Gapon wrote:
>> On 10/06/2019 06:07, Doug Moor
On Mon, 10 Jun 2019, Doug Moore wrote:
Log:
There are times when a len==0 parameter to mmap is okay. But on a
32-bit machine, a len parameter just a few bytes short of 4G, rounded
up to a page boundary and hitting zero then, is not okay. Return
failure in that case.
Some overflows still oc
-fwrapv concerns signed arithmetic. This calculation is with unsigned
arithmetic, and the possibility of wrapping around to 0 is part of the
language.
Doug Moore
On 6/10/19 1:35 AM, Andriy Gapon wrote:
> On 10/06/2019 06:07, Doug Moore wrote:
>> Author: dougm
>> Date: Mon Jun 10 03:07:10 2019
>>
On 10/06/2019 06:07, Doug Moore wrote:
> Author: dougm
> Date: Mon Jun 10 03:07:10 2019
> New Revision: 348843
> URL: https://svnweb.freebsd.org/changeset/base/348843
>
> Log:
> There are times when a len==0 parameter to mmap is okay. But on a
> 32-bit machine, a len parameter just a few bytes
I understand that I ought to have included this in the previous checkin.
MFC after: 3 days
and that my description of the appropriateness of len==0 passed to mmap
was imprecise. I regret the error.
Doug Moore
On 6/9/19 10:07 PM, Doug Moore wrote:
> Author: dougm
> Date: Mon Jun 10 03:07:10 201
Author: dougm
Date: Mon Jun 10 03:07:10 2019
New Revision: 348843
URL: https://svnweb.freebsd.org/changeset/base/348843
Log:
There are times when a len==0 parameter to mmap is okay. But on a
32-bit machine, a len parameter just a few bytes short of 4G, rounded
up to a page boundary and hitti
13 matches
Mail list logo
