RFR: 8273635: Attempting to acquire lock StackWatermark_lock/9 out of order with lock tty_lock/3

[David Holmes]

On Mon, 13 Sep 2021 20:04:56 GMT, Coleen Phillimore <coleenp at openjdk.org> wrote:

> This change reverts the rank ordering of ttyLock and StackWatermark_lock because the latter is held through a very large region and printing all of this to a buffer with xmlstream is non-trivial.
> With this change, if tty->print_cr() is done while holding the stackwatermark lock or lower (which is service ranking, etc), a lock inversion will happen with ttyLock.  This doesn't happen now because all the code in GC and much of the rest of the runtime use UL and not tty->print().
> Tested with tier1-6.

Hi Coleen,

The ranking restoration seems fine. One possible typo below.

Thanks,
David

test/hotspot/jtreg/compiler/uncommontrap/TestDeoptOOM.java line 39:

> 37:  * @test
> 38:  * @bug 8273456
> 39:  * @summary Test that ttyLock is ranked about StackWatermark_lock

s/about/above/ ?

-------------

Marked as reviewed by dholmes (Reviewer).

PR: https://git.openjdk.java.net/jdk/pull/5499