[ECOS] Re: redboot, slow upload, load -v -m ymodem failures for >150k binaries

[ECOS] Re: redboot, slow upload, load -v -m ymodem failures for >150k binaries

[Frank Lin]

Latter is an archive mail, now I got the same problem, anyone got resolution on it?


----- Original Message ----- 
a.. From: ecos at forumakad dot pl 
a.. To: ecos-discuss at ecos dot sourceware dot org 
a.. Date: Sun, 20 Feb 2005 15:18:17 +0100 
a.. Subject: [ECOS] redboot, slow upload, load -v -m ymodem failures for >150k binaries 


> Hello, 
> I'm trying to prepare application for EB42 board, unfortunatelly using ROM
> redboot I found that it work way slower than it's RAM counterpart ( I had
> to change speed from 38400 to 9600 to be able to upload anything ).
> Changing speed to 9600 fixed the problem for a while, but after I compiled
> in few options (lwip for example) my binaries grew in size, and now they're
> ~170-230k. That wouldn't be a big problem, because the board has 256k RAM,
> but I'm unable to upload those binaries to the board using RedBoot, errors
> appear after 160-200k of sent data ( using ymodem method ).
> 
> 
> This is extreme example, here's hello world:
> [ymodem upload - Press CTRL-C to quit]             
> Sending: hello.srec                                           
> Bytes Sent: 161664   BPS:765                                  
> RedBoot>  Transfer complete                                             
> RedBoot>                                                                
> Here's the same code with stack +4096 bytes larger:
> arm-elf-size hello.srec:
> text    data     bss     dec     hex filename
>       0   53860       0   53860    d264 hello.srec
> it's uploading fine for a few minutes:
> Sending: hello.srec                                           
> Ymodem sectors/kbytes sent: 615/76k   
> but then, it goes: 
> Retry 0: NAK on sector                                        
> Retry 0: Got 29 for sector ACK                                
> RedBoot>  Retry 0: NAK on sector                                        
> ** Error: Retry 0: Got 44 for sector ACK                                
> RedBoot>  Retry 0: NAK on sector                                        
> RedBoot>  Retry 0: Got 44 for sector ACK   
> 
> It happens every time with 'larger' binaries, it happens once in a while
> for smaller ones, but then I just retry and all is well.
> 
> When I revert to Angel boot loader, and upload binaries using arm-elf-gdb,
> all is well, and I can upload even ~200k execs, but then when I try to run
> them they just stand still.
> 
> What should I do? Is there some known problem with redboot? Should I
> attempt to fix it or maybe just fall back on angel? 
> What should I do to be able to create angel/standalone eCos binaries?
> 
> -- 
> Dariush Pietrzak,
>

Re: [ECOS] Re: redboot, slow upload, load -v -m ymodem failures for >150k binaries

[Jürgen Lambrecht]

Frank Lin wrote:

>Latter is an archive mail, now I got the same problem, anyone got resolution on it?
>
>
>----- Original Message ----- 
>a.. From: ecos at forumakad dot pl 
>a.. To: ecos-discuss at ecos dot sourceware dot org 
>a.. Date: Sun, 20 Feb 2005 15:18:17 +0100 
>a.. Subject: [ECOS] redboot, slow upload, load -v -m ymodem failures for >150k binaries 
>
>
>  
>
>>Hello, 
>>I'm trying to prepare application for EB42 board, unfortunatelly using ROM
>>redboot I found that it work way slower than it's RAM counterpart ( I had
>>to change speed from 38400 to 9600 to be able to upload anything ).
>>Changing speed to 9600 fixed the problem for a while, but after I compiled
>>in few options (lwip for example) my binaries grew in size, and now they're
>>~170-230k. That wouldn't be a big problem, because the board has 256k RAM,
>>    
>>
Be carefull with the RAM that RedBoot uses. The flash functions do
reside in RAM. (I don't know how big they are)
Regards,
J篓鹿rgen

>>but I'm unable to upload those binaries to the board using RedBoot, errors
>>appear after 160-200k of sent data ( using ymodem method ).
>>
>>
>>This is extreme example, here's hello world:
>>[ymodem upload - Press CTRL-C to quit]             
>>Sending: hello.srec                                           
>>Bytes Sent: 161664   BPS:765                                  
>>RedBoot>  Transfer complete                                             
>>RedBoot>                                                                
>>Here's the same code with stack +4096 bytes larger:
>>arm-elf-size hello.srec:
>>text    data     bss     dec     hex filename
>>      0   53860       0   53860    d264 hello.srec
>>it's uploading fine for a few minutes:
>>Sending: hello.srec                                           
>>Ymodem sectors/kbytes sent: 615/76k   
>>but then, it goes: 
>>Retry 0: NAK on sector                                        
>>Retry 0: Got 29 for sector ACK                                
>>RedBoot>  Retry 0: NAK on sector                                        
>>** Error: Retry 0: Got 44 for sector ACK                                
>>RedBoot>  Retry 0: NAK on sector                                        
>>RedBoot>  Retry 0: Got 44 for sector ACK   
>>
>>It happens every time with 'larger' binaries, it happens once in a while
>>for smaller ones, but then I just retry and all is well.
>>
>>When I revert to Angel boot loader, and upload binaries using arm-elf-gdb,
>>all is well, and I can upload even ~200k execs, but then when I try to run
>>them they just stand still.
>>
>>What should I do? Is there some known problem with redboot? Should I
>>attempt to fix it or maybe just fall back on angel? 
>>What should I do to be able to create angel/standalone eCos binaries?
>>
>>-- 
>>Dariush Pietrzak,
>>    
>>
>>
>

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Re: redboot, slow upload, load -v -m ymodem failures for >150k binaries

[ecos]

>If you have enought RAM spare, you could build a ROMRAM redboot. This
>lives in ROM, but at startup copies itself into RAM. It will then run
>much faster.
 Thanks for the tip, unfortunatelly EB42 is very limited when it comes to
RAM, there's 256k of it there, and as I heard this is minimal amount needed
for lwip+ppp.
 Also I don't think speed is the problem here ( but maybe I don't
understand something ), because smaller binaries upload just fine ( as I
experimentally found 156k binaries ( where 156k is .srec size ) upload fine,
while 160k cause countless problems ).

 One another thing - I don't understand why Angel works fine, while AFAIK
it's running from flash.
-- 
Dariush Pietrzak,
Key fingerprint = 40D0 9FFB 9939 7320 8294  05E0 BCC7 02C4 75CC 50D9

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss