On Win32 (ActivePerl 817, based on perl-5.8.8), there
are a few failures using Apache/2.2.2 with the svn
mod_perl sources. One set of failures is the
apr-ext tests:
=======================================================================
Failed Test Stat Wstat Total Fail Failed List of Failed
-----------------------------------------------------------------------------
t\apr-ext\brigade.t 5 1280 14 28 200.00% 1-14
t\apr-ext\bucket.t 5 1280 21 42 200.00% 1-21
t\apr-ext\finfo.t 5 1280 27 54 200.00% 1-27
t\apr-ext\perlio.t 5 1280 ?? ?? % ??
t\apr-ext\pool.t 5 1280 75 150 200.00% 1-75
t\apr-ext\table.t 5 1280 58 116 200.00% 1-58
t\apr-ext\threadmutex.t 5 1280 5 10 200.00% 1-5
t\apr-ext\uri.t 5 1280 36 72 200.00% 1-36
Failed 8/16 test scripts, 50.00% okay. 236/269 subtests
failed, 12.27% okay.
=======================================================================
These all relate to the same problem: a trace is
=======================================================================
apr_pool_create_ex(apr_pool_t * * 0x0140fd40,
apr_pool_t * 0x00000000,
int (int)* 0x00000000,
apr_allocator_t * 0x00000000) line 798
mpxs_apr_pool_create(interpreter * 0x002243b4,
sv * 0x0255abac) line 165
XS_APR__Pool_new(interpreter * 0x002243b4,
cv * 0x0255a3e4) line 245 + 43 bytes
=======================================================================
The apr_pool_create_ex() trouble on line 798 refers to
httpd-2.2.2/srclib/apr/memory/unix/apr_pools.c, which
refers to
============================================================
APR_DECLARE(apr_status_t) apr_pool_create_ex(
apr_pool_t **newpool,
apr_pool_t *parent,
apr_abortfunc_t abort_fn,
apr_allocator_t *allocator) {
apr_pool_t *pool;
apr_memnode_t *node;
*newpool = NULL;
if (!parent)
parent = global_pool;
if (!abort_fn && parent)
abort_fn = parent->abort_fn;
if (allocator == NULL)
allocator = parent->allocator; // line 798
=======================================================================
with the trouble being that parent is NULL.
The mpxs_apr_pool_create() function is found in
xs/APR/Pool/APR__Pool.h, while XS_APR__Pool_new() is
found in WrapXS/APR/Pool/Pool.c.
There's no analagous problems with the corresponding
apr/ tests (which use mod_perl.so), so it's unique
to using apr out of the mod_perl.so environment.
Has anyone, in a threaded or non-threaded environment,
seen any such problems with a unix flavour with Apache/2.2?
Does something in the above trigger any suggestions to try?
Thanks.
--
best regards,
Randy
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
