Re: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-15 Thread Mike Stump
On Feb 15, 2013, at 11:44 AM, Chao-Ying Fu wrote: > I know this, but it's very hard to educate people and let people admit it's > not CPU's fault, > when same software can run on other architectures. Show us the link were you tried. Some of might be happy to help beat on them. If you've not t

RE: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-15 Thread Chao-Ying Fu
Maciej W. Rozycki wrote: > > On Thu, 14 Feb 2013, Richard Sandiford wrote: > > > What about 64-bit targets? We can sometimes access doubles > using GPRs, > > so on 64-bit targets we could end up using LD and SD to > access a double > > even when this option disables LDC1 and SDC1. I think we'

Re: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-15 Thread Ralf Baechle
On Fri, Feb 15, 2013 at 01:23:14AM +, Maciej W. Rozycki wrote: > > What about 64-bit targets? We can sometimes access doubles using GPRs, > > so on 64-bit targets we could end up using LD and SD to access a double > > even when this option disables LDC1 and SDC1. I think we'd need to > > pat

Re: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-14 Thread Andrew Pinski
On Thu, Feb 14, 2013 at 5:23 PM, Maciej W. Rozycki wrote: > On Thu, 14 Feb 2013, Richard Sandiford wrote: > >> What about 64-bit targets? We can sometimes access doubles using GPRs, >> so on 64-bit targets we could end up using LD and SD to access a double >> even when this option disables LDC1 a

Re: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-14 Thread Maciej W. Rozycki
On Thu, 14 Feb 2013, Richard Sandiford wrote: > What about 64-bit targets? We can sometimes access doubles using GPRs, > so on 64-bit targets we could end up using LD and SD to access a double > even when this option disables LDC1 and SDC1. I think we'd need to > patch the move patterns as well.

RE: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-14 Thread Chao-Ying Fu
Richard Sandiford wrote: > Chao-Ying Fu writes: > > Hello All, > > > > Once in a while we got reports about programs (ex: > WebKit, FireFox) > > crash due to ldc1/sdc1 unaligned accesses on MIPS targets. The root > > cause is that programmers neglect the alignment issue and cast > > arbitrary

Re: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-14 Thread Richard Sandiford
Chao-Ying Fu writes: > Hello All, > > Once in a while we got reports about programs (ex: WebKit, FireFox) > crash due to ldc1/sdc1 unaligned accesses on MIPS targets. The root > cause is that programmers neglect the alignment issue and cast > arbitrary pointers to point to double variables. > >

Re: RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-13 Thread Andrew Pinski
On Wed, Feb 13, 2013 at 2:14 PM, Chao-Ying Fu wrote: > Hello All, > > Once in a while we got reports about programs (ex: WebKit, FireFox) > crash due to ldc1/sdc1 unaligned accesses on MIPS targets. The root cause is > that programmers > neglect the alignment issue and cast arbitrary pointers

RFC: [MIPS] Add an option to disable ldc1/sdc1

2013-02-13 Thread Chao-Ying Fu
Hello All, Once in a while we got reports about programs (ex: WebKit, FireFox) crash due to ldc1/sdc1 unaligned accesses on MIPS targets. The root cause is that programmers neglect the alignment issue and cast arbitrary pointers to point to double variables. Although the correct solution i