FW: is this a bug ?

[Andrew Haley]

On 10/16/19 10:22 AM, Aleksey Shipilev wrote:
> That would be weird.
> 
> The fixed test accesses DBB at offset 0, so the underlying storage (returned by
> Unsafe.allocateMemory) has to be unaligned in that case.

It'd be aligned because Unsafe_AllocateMemory0 uses malloc().
malloc() is aligned, by definition.

> This fixes the testbug and gets the proper behavior, apparently?
> 
> But then ByteBufferViewsAtomicityTests should crash the same way,
> no? And that would be odd, as JVM crash is not the expected behavior
> in this case.

I didn't read about a JVM crash.

> It surely is understandable, but not expected.

-- 
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671