--- In firebird-support@yahoogroups.com, "rtbunker" <rbunker.tsi@...>
wrote:

>
> --- In firebird-support@yahoogroups.com, Dmitry Yemanov <dimitr@>
> wrote:
> >
> > Could you please show us the complete stack trace?
> >
> >
> > Dmitry
> >
>
> Here's an example of stacks that are not in stall. I generated

these

> by just grabbing stacks in Visual Studio. If there's a firebird

tool

> to dump these or a better format, let me know, i'm not too familiar
> with the tracing facilities in firebird.
>
> Thread 1
> ----------
> Firebird::BePlusTree::getPrev(),
> Firebird::MemoryPool::internal_alloc() Line 1317 C++
> Firebird::MemoryPool::allocate_nothrow() Line 721 + 0x25 bytes

C++

> Firebird::AbstractString::reserveBuffer() Line 102 C++
> Firebird::AbstractString::baseAssign() Line 128 C++
> Firebird::AbstractString::vprintf() Line 403 + 0x16 bytes C++
> Firebird::AbstractString::printf(...) Line 358 C++
> CNTL_init() Line 82 + 0x13 bytes C++
> WinMain() Line 295 C++
> __tmainCRTStartup() Line 578 + 0x1d bytes
>
>
> Thread 2
> ----------
> CCH_release()
> CCH_handoff() Line 1624 + 0xd bytes C++
> BTR_find_page() Line 889 + 0x12 bytes C++
> BTR_evaluate() Line 797 C++
> pass1() Line 3866 + 0x1f bytes C+
>
>
> Thread 3
> ----------
> Firebird::MemoryPool::MemoryPool() Line 300 + 0x8c bytes C++
> Firebird::MemoryPool::internal_create() Line 1075 + 0xf bytes C++
> 7c901046 <-- jrd8_ddl shows a call into DYN_ddl, that must end
> up here?
> jrd8_ddl() Line 2217 C++
> isc_ddl() Line 1872 + 0x25 bytes C++
> DDL_execute() Line 298 + 0x1c bytes C++
> execute_request() Line 3345 + 0x6 bytes C
>
> The MemoryPool calls in Thread 3 are suspect to me as this would
> appear to be hanging in InitializeCriticalSection, which seems
> unlikely.
>
> Sometimes there are two calls into BePlusTree::getPrev on the stack.
> I wonder if that's where the culprit lies? Deadlock in multi-

threaded

> BePlusTree access?
> I consistently get 3 threads in some combination of freeBlock

access,

> CCH_release, and jrd8_ddl. I'll continue repro'ing and see if i get
> any more helpful stacks.
>
> :)ross
>

Hmm...i finally grabbed Windows symbols. That changes these stacks
considerably. Let me gather those and re-post