This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: class initialization check overhead
- From: Tom Tromey <tromey at redhat dot com>
- To: Adam Megacz <adam at megacz dot com>
- Cc: Jeff Sturm <jsturm at one-point dot com>, java at gcc dot gnu dot org
- Date: 25 Nov 2002 22:51:46 -0700
- Subject: Re: class initialization check overhead
- References: <Pine.LNX.4.44.0210231903100.4446-100000@ops2.one-point.com><86iszs3cyg.fsf@nowhere.com>
- Reply-to: tromey at redhat dot com
>>>>> "Adam" == Adam Megacz <adam@megacz.com> writes:
Adam> Could _Jv_InitClass be re-jiggered to perform all the register
Adam> saves itself (rather than asking the caller to do it)? In other
Adam> words, put a PUSHALL at the beginning and a POPALL at the end?
Is this the `fastcall' convention? We could do that on x86 machines.
You could see whether it helps pretty easily.
Adam> Or perhaps give gcj a list of classes that you know will be
Adam> referenced eventually, and have it run those <clinit>s
Adam> immediately upon startup.
As Kresten points out, this is tricky territory. The language
specification has a lot to say on this issue. Getting it
correct-but-slow has been difficult (I think we still have lurking
bugs...)
Tom