tree eb3916d9c0460cf0d96062489e1ec4ea7e7be01c
parent 78b0dfaba90874fe049194e9933f5b89b6158c1b
author Primiano Tucci <primiano@google.com> 1522321277 +0100
committer Primiano Tucci <primiano@google.com> 1522321277 +0100

Add retry logic to SharedMemoryABI::TryAcquireChunk

The current implementation was giving up immediately
when an AcquireForReading/Writing operation was failing
because producer and service are competing on the same page.
There was a TODO that was saying "this is too pessimistic"
alledging that was just an otpimization. However turned out
that was just incorrect. In fact when the service is asked
to move a chunk it cannot give up just because another chunk
in the same page was acquired, because it would cause a data
loss.

Bug: 77205968
Test: change default_page_layout=kPageDiv7, run perfetto_integrationtests --gtest_filter=PerfettoTest.VeryLargePackets
Change-Id: Idc43de60dc0dbe53ec7ba16fa754aa06de2c2ee7