Re: BUG #18259: Assertion in ExtendBufferedRelLocal() fails after no-space-left condition

Alexander Lakhin <exclusion@gmail.com> 于2023年12月27日周三 15:00写道：

Hello tender wang,

26.12.2023 19:55, tender wang write:

I tried to analyze the issue, and I found that it might be caused by this commit:

commit dad50f677c42de207168a3f08982ba23c9fc6720

       bufmgr: Acquire and clean victim buffer separately

Thanks for looking into it!

...

With debug logging added in this code within ExtendBufferedRelLocal():
        if (found)
        {
            BufferDesc *existing_hdr =
GetLocalBufferDescriptor(hresult->id);
            uint32      buf_state;

            UnpinLocalBuffer(BufferDescriptorGetBuffer(victim_buf_hdr));

            existing_hdr = GetLocalBufferDescriptor(hresult->id);
            PinLocalBuffer(existing_hdr, false);
            buffers[i] = BufferDescriptorGetBuffer(existing_hdr);

            buf_state = pg_atomic_read_u32(&existing_hdr->state);
            Assert(buf_state & BM_TAG_VALID);
            Assert(!(buf_state & BM_DIRTY));
            buf_state &= BM_VALID;
            pg_atomic_unlocked_write_u32(&existing_hdr->state, buf_state);
...
I see that it reached for the second INSERT (and NOSPC error) with
existing_hdr->state == 0x2040000, but for the third INSERT I observe
state == 0x0.

I wonder, if "buf_state &= BM_VALID" is a typo here, maybe it supposed to be
"buf_state &= ~BM_VALID" as in ExtendBufferedRelShared()...

 

Yeah, that's true.  I analyze this issue again, and I think the root cause is the " buf_state &= BM_VALID" .

In my report issue, buf_state & BM_VALID is true, but buf_state & BM_TAG_VALID is false. This situation is impossible.

It can't happen that the data in the local buffer pool is valid, but LocalBufHash has no entry.

I modified v1 patch, and attached v2 patch should fix the above issues.

Best regards,
Alexander