This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Unreviewed patches


On Mon, Jun 02, 2003 at 09:12:44PM -0700, H. J. Lu wrote:

On Mon, Jun 02, 2003 at 05:58:05PM -0700, Richard Henderson wrote:

> On Mon, Jun 02, 2003 at 12:40:40PM -0400, Jakub Jelinek wrote:

> > http://sources.redhat.com/ml/binutils/2003-05/msg00741.html

> > Since I think I suggested PT_GNU_STACK in the first place,
> I'm a bit biased, but I like this solution. It's the least
> amount of work for the kernel in execing a new application
> short of having a dedicated ET_FLAGS bit (which we don't).
> > The patch is ok.
>


I like the proposal. But I don't like the implementation. I'd like to see a generic note section for properties of a relocatable file. I will make a detailed suggestion tomorrow.


There are 2 separate things you argued about.
One is you wanted PT_GNU_PROPERTY segment instead of PT_GNU_STACK.
This is IMHO bad idea, since kernel/ld.so would need to dereference that
segment and parse its content to find out whether to use non-exec stack or
not.

Um, how is that significantly different to PT_GNU_STACK? The kernel/ld needs to check that section's contents for a "x".


Andrew


The other are the ET_REL-only sections used to set the default value.
Here, I don't understand if you want to keep the merged section in the
binary/libraries or not and why is a generic section with lots of flags
better than specialized section (I think ld has about the same amount
of work with both variants).

Jakub




Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]