Re: kernel .text and .rodata

[Masao Uebayashi <uebayasi%gmail.com@localhost>]

Who uses the _etext address?  Kernel itself?  Or some external loaders?

I thought most ports map .rodata as read, not read+executable.  It
seems most ports still map .rodata as read+executable?  Even if so,
excluding .rodata from the executable range might be beneficial for
ddb, profiler, or some cases.

On Wed, Nov 5, 2014 at 3:56 AM, Matt Thomas <matt%3am-software.com@localhost> wrote:
>
>> On Nov 4, 2014, at 7:53 AM, Dennis Ferguson <dennis.c.ferguson%gmail.com@localhost> wrote:
>>
>>
>> On 3 Nov, 2014, at 07:53 , Masao Uebayashi <uebayasi%gmail.com@localhost> wrote:
>>> After those loaders and other constrants are clarified, arm should set
>>> _etext after .test excluding .rodata.
>>
>> In an a.out binary _etext always marked the end of the stuff that could
>> be made read-only, whether text or data.  Is there another symbol that
>> is used for that now?
>
> In theory, __stub_end could be used for it since it's the last thing in .text