Post by thread:I solved it!! (was Help!! Power on, reset problems

Hi All. I solved my random error with power up and reset. It wasn't the watchdog (because I don't use it) It wasn't slow Vcc rise time. It was the Hi-tech C-compiler (Beta-3) who did it!! I discovered that when I used the "Full optimization-function" in the compiler, it didn't work. So a warning to everybody who use Hi-tech C-compiler (Beta-3), be careful when using "Full optimization". Thanks to everybody who tried to help me. Lars Johansson, Stockholm - Sweden. spam_OUTlrsTakeThisOuTstoa.mobile.telia.se (sorry for my english)

>It was the Hi-tech C-compiler (Beta-3) who did it!! >I discovered that when I used the "Full optimization-function" in >the compiler, it didn't work. So what was it about full optimization that did it? Did you look at the generated object code? What did it do differently? Andy ================================================================== Andy Kunz - Montana Design - 409 S 6th St - Phillipsburg, NJ 08865 Hardware & Software for Industry & R/C Hobbies "Go fast, turn right, and keep the wet side down!" ==================================================================

Hi all. È>It was the Hi-tech C-compiler (Beta-3) who did it!! È>I discovered that when I used the "Full optimization-function" in È>the compiler, it didn't work. È ÈSo what was it about full optimization that did it? Did you look at the Ègenerated object code? What did it do differently? È ÈAndy I'm not really sure, but maybe it have something to do with the code that saves away different register before/after interrupt. Unfortunally isn't my assembly skills so good (thats why I use C), so I send some code so you can see for yourself. --------------------------------------------------------------- NO OPTIMIZATION: saved_w ds 1 psect intcode bcf status,5 psect intret,class=CODE,delta=2 psect intret global int_restore int_restore movf saved_status,w movwf status swapf saved_w swapf saved_w,w retfie psect intentry,class=CODE,delta=2 psect intentry global int_entry int_entry movwf saved_w movf status,w movwf saved_status psect intsave,class=BANK0,space=1 psect intsave saved_status ds 1 psect intentry --------------------------------------------------------------------------- --------------------------------------------------------------------------- WITH "FULL OPTIMIZATION": global int_entry global int_restore psect intwsave,global,ovrld,class=BANK0,space=1 psect intret,global,class=CODE,delta=2 psect intentry,global,class=CODE,delta=2 psect intsave,global,class=BANK0,space=1 psect text2,local,class=CODE,delta=2 psect intwsave saved_w ds 1 psect intret int_restore movf saved_status,w movwf 3 swapf saved_w swapf saved_w,w retfie psect intentry int_entry movwf saved_w movf 3,w movwf saved_status psect intsave saved_status ds 1 psect text2 ------------------------------------------------------------------- //Bye Lars Johansson, Stockholm Sweden .....LRSKILLspam@spam@stoa.mobile.telia.se (sorry for my english)

In message <D221IXJ3MCFD*/R=A1/R=MRGST/U=LRS/@MHS.stoa.mobile.telia.se> PICLISTKILLspamMITVMA.MIT.EDU writes: > Hi All. > > I solved my random error with power up and reset. > It wasn't the watchdog (because I don't use it) > It wasn't slow Vcc rise time. > > It was the Hi-tech C-compiler (Beta-3) who did it!! > I discovered that when I used the "Full optimization-function" in > the compiler, it didn't work. > So a warning to everybody who use Hi-tech C-compiler (Beta-3), > be careful when using "Full optimization". > Any comments on this one, Clyde? Regards, Mike Watson

On Thu, Jul 03, 1997 at 10:46:53AM +0000, Mayes uk wrote: > > Any comments on this one, Clyde? No, Mike, not really. The beta is just that, a beta. It has bugs (though I have no idea if the behaviour reported is due to a compiler bug or not). No similar problem has been reported with the released version. Clyde -- Clyde Smith-Stubbs |HI-TECH Software, |Email: .....clydeKILLspam.....htsoft.com Ph: +61 7 3354 2411|P.O. Box 103, Alderley,|WWW: http://www.htsoft.com/ Fax: +61 7 3354 2422|QLD, 4051, AUSTRALIA. |PGP: finger EraseMEclydespam_OUTTakeThisOuThtsoft.com --------------------------------------------------------------------------- ANSI C for the PIC! Now shipping! See http://www.htsoft.com for more info.

At 21:31 03/07/97 +1000, Clyde Smith-Stubbs wrote: >On Thu, Jul 03, 1997 at 10:46:53AM +0000, Mayes uk wrote: >> >> Any comments on this one, Clyde? > >No, Mike, not really. The beta is just that, a beta. It has bugs (though I >have no idea if the behaviour reported is due to a compiler bug or not). >No similar problem has been reported with the released version. Still considering your compiler as an option, I sure am interested to know whether this specific problem is a bug, and whether this bug has been resolved in the final version. Until you know what happened, you can't tell, and I thought you might be interested in such stuff. I sure am. ------------------------------------- Gerhard Fiedler <gerhardspam_OUTpobox.com> S‹o Paulo - Brazil

On Fri, Jul 04, 1997 at 04:12:21PM +0000, Gerhard Fiedler wrote: > > Still considering your compiler as an option, I sure am interested to know > whether this specific problem is a bug, and whether this bug has been > resolved in the final version. Until you know what happened, you can't > tell, and I thought you might be interested in such stuff. I sure am. Yes, I'm interested, but without more information, it's impossible to tell. All bug reports we get, whether related to the beta or released versions, are looked at. But so far I have only seen some out-of-context code fragments (don't even know which chip is being used). My feeling at this time is that the behaviour is coincidental, but there's not enough information to tell. Any bug reports should be sent to @spam@picKILLspamhtsoft.com. This is not really the right place. Clyde -- Clyde Smith-Stubbs |HI-TECH Software, |Email: KILLspamclydeKILLspamhtsoft.com Ph: +61 7 3354 2411|P.O. Box 103, Alderley,|WWW: http://www.htsoft.com/ Fax: +61 7 3354 2422|QLD, 4051, AUSTRALIA. |PGP: finger RemoveMEclydeTakeThisOuThtsoft.com --------------------------------------------------------------------------- ANSI C for the PIC! Now shipping! See http://www.htsoft.com for more info.

At 08:46 05/07/97 +1000, Clyde Smith-Stubbs wrote: >Any bug reports should be sent to spamBeGonepicspamBeGonehtsoft.com. This is not really the >right place. If it is something that appears (or appeared) in the full version, I think this _is_ the right place. There's of course no better place than this list for _all_ bug reports concerning _all_ PIC compilers (whether fixed afterwards or not), don't you think? :-) ------------------------------------- Gerhard Fiedler <TakeThisOuTgerhardEraseMEspam_OUTpobox.com> S‹o Paulo - Brazil

On Sat, Jul 05, 1997 at 08:03:56PM +0000, Gerhard Fiedler wrote: > At 08:46 05/07/97 +1000, Clyde Smith-Stubbs wrote: > >Any bug reports should be sent to RemoveMEpicTakeThisOuThtsoft.com. This is not really the > >right place. > > If it is something that appears (or appeared) in the full version, I think > this _is_ the right place. There's of course no better place than this list > for _all_ bug reports concerning _all_ PIC compilers (whether fixed > afterwards or not), don't you think? :-) No, I don't. The reason is that there's no guarantee that the bug report will get to us, list mail is not 100% reliable. I think this thread has taken up far too much space on the list, so perhaps we can put an end to it. Here are the facts as I have seen them on the list: 1) Someone posted a message saying their program randomly failed to come up after reset; 2) That person then posted a message saying they were using the Beta3 version of our compiler, and that turning OFF optimization cured the problem; 3) Some code fragments were posted showing the interrupt save/restore code with and without optimization - but the two pieces of code were identical. 4) No other information has been provided on the list - not even the particular chip involved. 5) No contact has been made directly with us about it. The above does not constitute a bug report, and has no bearing on the released version of the software. If anyone has a bug report related to the beta 3 version then we are very interested to hear about it - we will verify if it still exists. We can't do that with the above information because it does show anything. -- Clyde Smith-Stubbs |HI-TECH Software, |Email: clydeEraseME.....htsoft.com Ph: +61 7 3354 2411|P.O. Box 103, Alderley,|WWW: http://www.htsoft.com/ Fax: +61 7 3354 2422|QLD, 4051, AUSTRALIA. |PGP: finger EraseMEclydehtsoft.com --------------------------------------------------------------------------- ANSI C for the PIC! Now shipping! See http://www.htsoft.com for more info.