[PATCH 2/2] ldso: make hashcode handling more generic

[Chris Zankel]

Hi Bernhard,

On 1/22/14, 2:15 PM, Bernhard Reutner-Fischer wrote:
> On 13 January 2014 17:18, Chris Zankel <chris at zankel.net> wrote:
>> Hi,
>>
>> I was wondering if anyone had a chance to look at these changes again? I
>> have attached the two patches we propose so we can add support for TLS
>> descriptors.
>>
>> It would be great to get an ACK for it (or NACK with maybe some other
>> suggestions).
> I did not have a chance to look at it yet, no.
> I've been side-tracked attempting to understand gold's POV on TLS (or
> static tls or however it is called, think dreaded tsd. Holler if that
> rings a bell,
Do you have any pointers to it? I'm not sure I know who or what 'gold' 
is or means?
I'd be curious, though...
> i could use somebody filling in my gaps of interest in that area ;).. Going
> through the backlog now but it's getting too late for me today to continue.
No worries. I'm just glad you haven't forgotten :-)

Regarding those patches, they should  be transparent for all other 
architectures, ie. not change anything, but are a requisite for to add 
TLS/NPTL support for Xtensa. For Xtensa, we have to use what's called 
TLS descriptors, and which will come in additional patches.

The two changes in that patch are:
- change struct funcdesc_value** to void**, which I think is save,
- pass the two functions, hash_pointer and eq_pointer as additional 
arguments to htab_find_slot(), but any advanced compiler should just 
inline them again.

(Note that the entire patch series for Xtensa is under 
https://github.com/foss-xtensa/uClibc.git in the xtensa_nptl branch)

Thanks,
-Chris