Hi,

a snippet of my c-code looks like this:

----------------------------------------------------------------------------------------------------------------
if (tempa == 0)
{
Ergebnis = 0;
}
}
Wait_ms (100);
----------------------------------------------------------------------------------------------------------------

With a breakpoint on the if statement the contents of tempa is 9 (= correct in this case).
Next singlestep stops @ Ergebnis = 0 which pretends that tempa is 0 (= obviously not correct).
With a look at the assembly the program flow is ok because with the BNE in effect program execution
resumes with the preparation of the wait-loop. But this is not part of the Ergebnis statement.


The program is running fine but debugging is misleading .

----------------------------------------------------------------------------------------------------------------

In assembly the code reads :


if (tempa == 0) ;Breakpoint, tempa currently = 9
080039AE 7800 LDRB R0, [R0, #0]
080039B0 2800 CMP R0, #0
080039B2 D102 BNE 0x080039BA ;Branch active because tempa != 0
080039B4 A802 ADD R0, SP, #8
080039B6 2100 MOVS R1, #0
{
Ergebnis = 0;
080039B8 7001 STRB R1, [R0, #0]
080039BA 2064 MOVS R0, #100 ;Branch destination = start of Wait_ms(100)
}
}
Wait_ms (100);
080039BC F7FE FC70 BL Wait_ms
080039C0 A802 ADD R0, SP, #8

Hello Nino,

thanks for your reply.

Ozone-Version is 3.22.
The code snippet shows just some lines of a complete (well-running) function with about 80 lines.
I compile it with Keil-ARM , load the axf with Keil+JLink+SWD into the STM32F051, debug the named lines with Keil and stepping is ok.
For some reason Ozone is not able to download this axf so I have to "Attach to running program", but stepping is ok, too.
(JLink reports an unspecified download error)
Using the same sourcefiles in emStudio for ARM 5.30, compiling with gcc (hopefully the most recent version), loading the gcc-axf with
Ozone (download running fine) , stepping in Ozone (or emStudio) runs into "Ergebnis = 0".

The Keil-Disassembly looks like this:

76: if (tempa == 0)
77: {
0x08002FA8 2D00 CMP r5,#0x00
0x08002FAA D100 BNE 0x08002FAE
78: Ergebnis = 0;
79: }
80: }
81: //-----------------------------------------------------------------------------
0x08002FAC 2400 MOVS r4,#0x00
82: Wait_ms (100); // SR macht ca 50ms Driver-OFF nach letzter Sendung
83: //-----------------------------------------------------------------------------
0x08002FAE 2064 MOVS r0,#0x64
0x08002FB0 F000FD28 BL.W Wait_ms (0x08003A04)
84: if (Ergebnis == 0)

I could send a small video, but it is 1.6MBytes long.

Best regards
Detlef

I wrote some senseless lines to get closer to the problem (see attachment). Mind you, everything is working fine.
But you're upset when single stepping shows the wrong statement.

Hi Nino,

with today's installation of emStudio 5.40 the Problem is solved.
Please see the attached screenshot :
The disassembly shows the same compilation, only the blockmarkers for lines 53 and 55 are placed one line earlier.

Best regards
Detlef

Hello Nino,


the problem occured while testing a project with Ozone which had been edited and compiled in emStudio, so I started with focus on Ozone.
It is an old Keil project which I imported into emStudio, so I had the chance to compare Ozone's reactions with these two compilations.

And yes, the issue has been resolved by updating emStudio to version 5.40 which might have included an update of the compiler, too.

Best regards,
Detlef