[PATCH v6 13/17] arm64:ilp32: add sys_ilp32.c and a separate table (in entry.S) to use it

[cmetcalf@ezchip.com (Chris Metcalf)]

On 11/15/2015 10:18 AM, Arnd Bergmann wrote:
> On Friday 13 November 2015 17:10:44 Arnd Bergmann wrote:
>> On Friday 13 November 2015 07:38:49 Andrew Pinski wrote:
>>> On Fri, Nov 13, 2015 at 7:34 AM, Arnd Bergmann <arnd@arndb.de> wrote:
>>>> On Thursday 12 November 2015 14:47:18 Andreas Schwab wrote:
>>>>> Arnd Bergmann <arnd@arndb.de> writes:
>>>>>
>>>>>> On Thursday 12 November 2015 10:44:55 Andreas Schwab wrote:
>>>>>>> Arnd Bergmann <arnd@arndb.de> writes:
>>>>>>>
>>>>>>>> What do you mean with 32-bit off_t?
>>>>>>> An ABI with 32-bit off_t, ie. all currently implemented 32-bit ABIs.
>>>>>>>
>>>>>>>> Do you mean that glibc emulates a 32-bit off_t on top of the 64-bit
>>>>>>>> __kernel_loff_t?
>>>>>>> Glibc is bridging the user-space ABI to the kernel ABI.
>>>>>> Ok, but why?
>>>>> That's how the ABI is defined right now.  I didn't make that up.
>>>> Ok, I guess it will remain a mystery then.
>>> The biggest question is here is how much compatibility do we want with
>>> other 32bit ABIs?
>>> Do we want off_t to be 32bit or 64bit?
>> I would much prefer off_t to be defined as __kernel_loff_t unconditionally,
>> with no support for _FILE_OFFSET_BITS == 32. This is at least what I had
>> in mind when I wrote the asm-generic/unistd.h header.
>>
>> We should probably find out what happened for the other glibc ports that
>> were implemented for the architectures using this. It's possible that
>> there was a good reason for supporting _FILE_OFFSET_BITS == 32 at the
>> time, but I can't think of one and maybe it is one that is no longer
>> valid.
>>
>> Do you know what x86/x32 does for off_t? Do they also implement both
>> _FILE_OFFSET_BITS == 32 and _FILE_OFFSET_BITS == 64 on top of the
>> 64-bit __kernel_off_t?
> I just did a little bit of digging through glibc history and found that
> Chris Metcalf added the files that are now in
> sysdeps/unix/sysv/linux/generic/wordsize-32/ and that provide the
> implementation for 32-bit off_t in glibc on top of the 64-bit
> __kernel_off_t.
>
> Chris, do you remember what led to that? Do you think we still need
> to have 32-bit off_t on all new architectures, or could we move
> on to making 64-bit off_t the default when adding a port?

I think there are two questions here.  The first is whether glibc will change the
default for _FILE_OFFSET_BITS to be 64.  This has been discussed in the past, e.g.:

https://sourceware.org/ml/libc-alpha/2014-03/msg00351.html

I've added Rich, Paul, Joseph, and Mike to the cc's as they are probably a good
subset of libc-alpha to help comment on these issues.  My sense is that right now,
it wouldn't be possible to add a 32-bit architecture with a non-32-bit default
for _FILE_OFFSET_BITS.  And, obviously, this is why, when I added the tilegx32
APIs to glibc in 2011, I needed to provide _FILE_OFFSET_BITS=32 support.

As to the kernel APIs, certainly tilegx32 only has the stat64 API; I just arranged
that the userspace structures are file-offset-bits-agnostic by using ifdefs to
either put a 64-bit value or a (32-bit-value, 32-bit-pad) in the structure.
See sysdeps/unix/sysv/linux/generic/bits/stat.h for example.  While the
__field64() macro is kind of nasty, it does provide the 32-bit off_t to those
programs that want it without any particular cost elsewhere in the code.

-- 
Chris Metcalf, EZChip Semiconductor
http://www.ezchip.com