Re: Naive Long Branch question
David Mosberger <[email protected]> Thu, 17 Jul 2003 10:46:57 -0700
| Newsgroups | gmane.linux.redhat.ia64.general |
|---|---|
| Message-ID | <[email protected]> |
>>>>> On Wed, 16 Jul 2003 21:40:06 -0700 (PDT), Ravi g <[email protected]> said: Ravi> Hi everybody, Is Long Branch Operation supported in Linux. Yes. Ravi> Ifso from which version of the kernel. I don't remember the exact revision, but it's been there for a long time. Ravi> How does it support it. If i were to write directly a Ravi> LongBranch instruction to memory with my own compiler does the Ravi> Operating System have the fault handler to catch that and Ravi> emulate it. On Itanium (Merced), yes. On Itanium 2 (McKinley or Madison), the hardware will execute it directly. Ravi> And in case of my own compiler who will set the 1b bit of Ravi> CPUID register 4,which is needed to the presence of such an Ravi> instruction. You don't. That bit is defined by hardware. The branch-long emulation is there for software which doesn't check that bit. --david