Re: [AVR] Fix target/34888

2011-08-04 Thread Georg-Johann Lay
Richard Henderson wrote: > On 08/04/2011 09:04 AM, Denis Chertykov wrote: >> Thank you for explanation. >> I have a very clean understanding of whole picture. >> May be better to use define_peephole2 with 3 insns as input and 1 as >> output for easy understanding. > > *shrug* Maybe. Then you also

Re: [AVR] Fix target/34888

2011-08-04 Thread Denis Chertykov
2011/8/4 Richard Henderson : > On 08/04/2011 09:04 AM, Denis Chertykov wrote: >> Thank you for explanation. >> I have a very clean understanding of whole picture. >> May be better to use define_peephole2 with 3 insns as input and 1 as >> output for easy understanding. > > *shrug* Maybe.  Then you a

Re: [AVR] Fix target/34888

2011-08-04 Thread Richard Henderson
On 08/04/2011 09:04 AM, Denis Chertykov wrote: > Thank you for explanation. > I have a very clean understanding of whole picture. > May be better to use define_peephole2 with 3 insns as input and 1 as > output for easy understanding. *shrug* Maybe. Then you also have to check for whether TMP is d

Re: [AVR] Fix target/34888

2011-08-04 Thread Denis Chertykov
2011/8/4 Richard Henderson : > On 08/03/2011 11:09 PM, Denis Chertykov wrote: >> 2011/8/4 Richard Henderson : >>> When a frame pointer is in use, we can optimize popping all >>> queued parameters via a simple move from the frame pointer >>> instead of an addition to the stack pointer. >>> >>> The n

Re: [AVR] Fix target/34888

2011-08-04 Thread Richard Henderson
On 08/03/2011 11:09 PM, Denis Chertykov wrote: > 2011/8/4 Richard Henderson : >> When a frame pointer is in use, we can optimize popping all >> queued parameters via a simple move from the frame pointer >> instead of an addition to the stack pointer. >> >> The new sequence is 4 insns, the old seque

Re: [AVR] Fix target/34888

2011-08-04 Thread Georg-Johann Lay
Richard Henderson wrote: > When a frame pointer is in use, we can optimize popping all > queued parameters via a simple move from the frame pointer > instead of an addition to the stack pointer. > > The new sequence is 4 insns, the old sequence was 9 insns. > > Committed. > > r~ 4 insns is odd.

Re: [AVR] Fix target/34888

2011-08-03 Thread Denis Chertykov
2011/8/4 Richard Henderson : > When a frame pointer is in use, we can optimize popping all > queued parameters via a simple move from the frame pointer > instead of an addition to the stack pointer. > > The new sequence is 4 insns, the old sequence was 9 insns. > > Committed. It seems strange for

[AVR] Fix target/34888

2011-08-03 Thread Richard Henderson
When a frame pointer is in use, we can optimize popping all queued parameters via a simple move from the frame pointer instead of an addition to the stack pointer. The new sequence is 4 insns, the old sequence was 9 insns. Committed. r~ PR target/34888 * config/avr/avr.md: New s