Submitted by: @hvlad
Is related to CORE923
The bug was reported at https://www.sql.ru/forum/1333365/
The problem could happen when:
- the op_event packet is longer than the XNET transmission buffer length (100 bytes), and
- client start to read data a bit later than server write first 100 bytes.
In this case race between writer and reader could happen and reader could miss first 100 bytes of the packet contents.
The bug is close to impossible to reproduce without artificial manipulation by running threads using debugger.
It might explain why it was not detected so far, since XNET was introduced.
Commits: 4fd96ee 197615c
Submitted by: @hvlad
Is related to CORE923
The bug was reported at https://www.sql.ru/forum/1333365/
The problem could happen when:
- the op_event packet is longer than the XNET transmission buffer length (100 bytes), and
- client start to read data a bit later than server write first 100 bytes.
In this case race between writer and reader could happen and reader could miss first 100 bytes of the packet contents.
The bug is close to impossible to reproduce without artificial manipulation by running threads using debugger.
It might explain why it was not detected so far, since XNET was introduced.
Commits: 4fd96ee 197615c