Hi, Thanks for looking into this!
On Thu, 31 Oct 2024 at 21:18, Andrey M. Borodin <x4...@yandex-team.ru> wrote: > > > On 8 Aug 2024, at 11:32, Nazir Bilal Yavuz <byavu...@gmail.com> wrote: > > > > Any feedback would be appreciated. > > I've took a look into the patch. It seems to me that you add new block > numbers to the read stream until you have buffers. So when there are no more > buffers you will still have some queued blocks. > Maybe can you change the logic so that number of free buffers must be enough > to allocate all blocks in look-ahead distance? I see what you mean. When the have_free_buffer() function returns false in the callback, there are still queued blocks in the stream although there are no free buffers in the buffer pool. I think the best way to solve this is to get the number of free buffers in the buffer pool by 'BufferStrategyControl.lastFreeBuffer - BufferStrategyControl.firstFreeBuffer' and then compare it with 'stream->pending_read_nblocks'. When the 'stream->pending_read_nblocks == number_of_free_buffers_in_buffer_pool', end the stream. The problem with that is stream->pending_read_nblocks isn't public, also I am not sure whether 'BufferStrategyControl.lastFreeBuffer - BufferStrategyControl.firstFreeBuffer' is safe to use. -- Regards, Nazir Bilal Yavuz Microsoft
