RFC: __kernel_time_t

[Aurelien Jarno]

Robert Millan a écrit :
> On Tue, Jan 03, 2006 at 08:33:58PM +0100, Robert Millan wrote:
> 
>>On Tue, Jan 03, 2006 at 08:24:57PM +0100, Aurelien Jarno wrote:
>>
>>>On Tue, Jan 03, 2006 at 08:20:27PM +0100, Robert Millan wrote:
>>>
>>>>On Tue, Jan 03, 2006 at 08:07:54PM +0100, Aurelien Jarno wrote:
>>>>
>>>>>First, you will se that our definition is wrong in case of amd64, ia64
>>>>>and sparc64. Then __time_t in the kernel == __time_t in userland on all 
>>>>>platforms but alpha. That's probably because alpha is the first 64-bit
>>>>>port and at the time it was done, the wrong choice has been done about
>>>>>the type of __time_t.
>>>>
>>>>Maybe this helps:
>>>>
>>>>http://lists.alioth.debian.org/pipermail/glibc-bsd-devel/2004-December/000324.html
>>>>
>>>>According to Bruno, alpha's 32bit time_t won't overflow untill 2038.  I don't
>>>>think anybody will care about alpha at that time, so my suggestion would be to
>>>>just ignore the problem and use whatever the kernel uses.
>>>
>>>What do you mean by "ignoring the problem"? If we change nothing, that's
>>>mean we will keep our ugly-hack, and we will also have problems with the
>>>amd64 port.
>>
>>I mean getting rid of __kernel_time_t, of course.  That'll give you an ABI
>>change in alpha, but who cares :)
> 
> 
> Another option is to patch the kernel to define time_t as 64bit for alpha, then
> let the change happen in kernel abi instead.
> 
This would break compatibility with FreeBSD, ie we won't be able to 
execute FreeBSD binaries on GNU/kFreeBSD. I have already implemented 
compatibily wrappers in the SVN.

Anyway, I don't think somebody will start a GNU/kFreeBSD port on alpha. 
This is not a really common architecture those days.

Bye,
Aurelien

-- 
   .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
  : :' :  Debian developer           | Electrical Engineer
  `. `'   aurel32 at debian.org         | aurelien at aurel32.net
    `-    people.debian.org/~aurel32 | www.aurel32.net