From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Merrill To: Sylvain Pion Cc: egcs@cygnus.com Subject: Re: C++ default copy ctor not optimal Date: Sun, 14 Feb 1999 21:01:00 -0000 Message-id: In-reply-to: Sylvain Pion's message of Sun, 14 Feb 1999 19:57:43 +0100 References: <19990212120037.C13091@rigel.inria.fr> <19990212134607.F13091.cygnus.egcs@rigel.inria.fr> <19990214195743.A1576@rigel.inria.fr> X-SW-Source: 1999-02/msg00632.html >>>>> Sylvain Pion writes: > On Fri, Feb 12, 1999 at 01:41:03PM -0800, Jason Merrill wrote: >> This is a backend issue; the frontend just tells the backend "copy this >> struct". The insns used are up to the backend. > Ok, then is it possible to have the backend do some kind of memcpy using > the FP registers (this reminds me some old linux kernel patch never > included...) ? What are the problems with this approach ? The backend should be able to determine that the struct it's copying is composed of FP values and use FP instructions for the copy. I don't think we can do that in general on the x86 because the FPU faults if you try to load an invalid FP value, but I may be remembering wrong. Jason From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Merrill To: Sylvain Pion Cc: egcs@cygnus.com Subject: Re: C++ default copy ctor not optimal Date: Sun, 28 Feb 1999 22:53:00 -0000 Message-ID: References: <19990212120037.C13091@rigel.inria.fr> <19990212134607.F13091.cygnus.egcs@rigel.inria.fr> <19990214195743.A1576@rigel.inria.fr> X-SW-Source: 1999-02n/msg00627.html Message-ID: <19990228225300.9llBK6RTlEMAOBVWW-KAI1e1TbjgUqRVG6tLF5tVJks@z> >>>>> Sylvain Pion writes: > On Fri, Feb 12, 1999 at 01:41:03PM -0800, Jason Merrill wrote: >> This is a backend issue; the frontend just tells the backend "copy this >> struct". The insns used are up to the backend. > Ok, then is it possible to have the backend do some kind of memcpy using > the FP registers (this reminds me some old linux kernel patch never > included...) ? What are the problems with this approach ? The backend should be able to determine that the struct it's copying is composed of FP values and use FP instructions for the copy. I don't think we can do that in general on the x86 because the FPU faults if you try to load an invalid FP value, but I may be remembering wrong. Jason