[PATCH 2/2] Initialize tunable list with the GLIBC_TUNABLES environment variable

[Carlos O'Donell]

On 01/13/2016 03:38 AM, Andreas Schwab wrote:
> Siddhesh Poyarekar <sid@reserved-bit.com> writes:
> 
>> A single load (and constructor call) does not matter since there's
>> nothing to race against.  The case of concurrent constructor calls I
>> mentioned in the following paragraph; the load lock is sufficient to
>> protect it.
> 
> PR19448 is about removing that lock.

If the code has comments that it needs protection from load lock then
such code will be reviewed as we fix BZ19448. Which is why I suggested
comments to that extent. The removal of the load lock won't be easy
and will require quite a bit of review.

Cheers,
Carlos.