RTOS/TM4C123GH6PM: "remove()" works BUT rename() does not. How do I properly specify the arguments to rename?

Hi Charles,

The tokenizer, interpreter, RTOS and everything  runs in about 90K (so far) in the TM4C123GH6PM.

(It's all really quite simple. The 'language' is roughly modeled after a simple 'assembly language' (like for a 6502 or 6811) in which every instruction begins with  an 'opcode' verb. All the tokenizer does is look up the first word in each line and replace it with the index number (stringified) of the list in which the word is found. During interpretation each line read from a file is parsed into an array of tokens and the first token, the opcode, is converted by atoi() and fed into a switch statement. Each 'case' in the switch calls one subroutine which handles the intent of the instruction using the rest of the tokens as 'operands'. Interpretation of each file is managed via a 'virtual cpu' struct. Here's a clarifying picture...

Cheers,

Dave

Hello Sean,

About:

1. I don't know. I think it's not based on an SDK.

2. rename() is returning -1.

3 The issue is absolutely consistent.

4 In 'stdio.h' changed the _NFILE entry value from 10 to 15.

'file.h' changed the _NSTREAM entry value from 10 to 15.

I was advised to delete rtsv7M4_T_le_v4SPD16_eabi.lib to force a rebuild of it which would incorporate my changes, and so I did and it worked. _ftable now has 15 entries.

RemoveFile() works fine, I stepped thru it and saw where it returned 0.

I stepped thru renameFile(), couldn't make sense of most of the assembly (not familiar with ARM). I did see where it returned -1 somewhat after finddevice() returned null (as shown below).

Are there some files I should send you?

Thanks for your help.

Cheers,

Dave

p.s. Here is how my debug console looks after the robot boots and I tell it to tokenize file k.src

(things line up better on the debug screen than it appears here)...

Hello  World

Opened SYSTEM.TXT r
Closed SYSTEM.TXT
System restored

Clock started: Millisec

!ReportBump b 7 1
!ReportUnit unit 1_5E98C1
!ReportVolt batt 11.9
!ReportTime setTime 3760768312

Clock started: PID
Clock started: Location
Clock started: AtoD
Clock started: ArmServos
Clock started: Scanner

Ok: tokenize k
Tokenizing K.SRC
PASS 1 --------------------------------------
Opened K.SRC r
Opened K.BOB w
    OP OPERANDS
    74 on
    43 x 10
    52
    53 Secs 0 Msecs 0
    46 @2Seek Test\n
    55 ....... 0 x <
    54 ....... x 5 <=
    46 @%d@%d@6 x = %d\n,x,x,x
    81
    42 x
    56
    46 @2 x = %d\n,x
    46 @3This test took\n
    46 @4%d.%d\n, Secs , Msecs
    46 @5seconds.\n
    35
Closed K.SRC
Closed K.BOB

PASS 2 --------------------------------------
Opened K.BOB r
Opened TMP.BOB w
   POSITION OP OPERANDS
      0 74 on
      6 43 x 10
     14 52
     17 53 Secs 0 Msecs 0
     35 46 @2Seek Test\n
     52 55 125     0 x <
     69 54 117     x 5 <=
     87 46 @%d@%d@6 x = %d\n,x,x,x
    114 81
    117 42 x
    122 56
    125 46 @2 x = %d\n,x
    142 46 @3This test took\n
    164 46 @4%d.%d\n, Secs , Msecs
    191 46 @5seconds.\n
    207 35
Closed K.BOB
Closed TMP.BOB
Removed K.BOB
NOT RENAMED:TMP.BOB to K.BOB
retval = -1
Rename: Domain error
Tokenized K.SRC

Ok: list k.src C
Opened K.SRC r
    // Seek Test
    trace on
    var x 10
    cls
    set Secs 0 Msecs 0
    ? @2Seek Test\n
    while 0 < x
        if x <= 5
            ? @%d@%d@6 x = %d\n,x,x,x
        endif
        -- x
    wend
    ? @2 x = %d\n,x
    ? @3This test took\n
    ? @4%d.%d\n, Secs , Msecs
    ? @5seconds.\n
    end
Closed K.SRC
Done Listing

Ok:

Hi Sean,
I thought I'd let rename() swim around in the unconscious for a while as I worked out other problems in the system. Now everything works as it should (I believe everything ho ho , more testing in order natch). To enable this activity I wrote a 'work around' in which I simply copied the TMP.BOB file over the K.BOB file thereby achieving the same result as rename().

So just now...
I stepped thru rename() in assembler on one screen and observed lowlevel.c on another screen trying to find the condition causing rename to fail. Here are my notes copypasted from both screens...
___________________________________________________________________________________

void renameFile(char *old, char *new){ <--- my rename function
char oldbuf[20];
char newbuf[20];
sprintf(oldbuf, "fat:0:%s", old);
sprintf(newbuf, "fat:0:%s", new);
if(rename(oldbuf, newbuf) isNot 0){... <--- disassembly starts

rename():
getdevice():
finddevice():
return NULL;
dev = finddevice(devname); <--- dev is 'fat'

return stdevice; /* the "standard" device - host I/O */
old_dev = getdevice(&old_name); <--- old_dev is 'fat'
new_dev = getdevice(&new_name); <--- new_dev is 'fat'
if (old_dev != new_dev)
result = (*(old_dev->RENAME)) (old_name, new_name);

Can't find a source file at "/db/vtree/library/trees/zumaprod/zumaprod-f33/tirtos_tivac_2_12_01_33/products/bios_6_41_04_54/packages/ti/sysbios/fatfs/ffcio.c"
Locate the file or edit the source lookup path to include its location.
ffcio_rename():
00013186: B508 push {r3, lr}
00013188: F7F5FEF8 bl #0x8f7c

f_rename():
00008f7c: B570 push {r4, r5, r6, lr}
.
.
.
00008fc8: F005F938 bl #0xe23c
00008fcc: 1C04 adds r4, r0, #0
00008fce: D059 beq $C$L173
00008fd0: 2C04 cmp r4, #4
00008fd2: D15A bne $C$L175 <--- r4 is 6, not 0
$C$L175:
0000908a: 9801 ldr r0, [sp, #4]
0000908c: 4621 mov r1, r4
0000908e: F009FEA5 bl #0x12ddc
00009092: 4620 mov r0, r4
00009094: B018 add sp, #0x60
00009096: BD70 pop {r4, r5, r6, pc}

0001318c: 2800 cmp r0, #0 <--- r0 is 6
0001318e: BF14 ite ne
00013190: F04F30FF mov.w r0, #-1 <--- this executes
00013194: 2000 movs r0, #0
00013196: BD08 pop {r3, pc}
result = (*(old_dev->RENAME)) (old_name, new_name); <--- result is -1
00002bb2: 4628 mov r0, r5
00002bb4: BD3E pop {r1, r2, r3, r4, r5, pc}

result = (*(old_dev->RENAME)) (old_name, new_name); <--- result is -1
__TI_resource_unlock(__TI_LOCK_DEVICE_TBL);

0000d95a: B928 cbnz r0, #0xd968 <--- r0 is -1
104 printB("RENAME %s \rto %s \rFAILED\n", oldbuf, newbuf); <--- #0xd968 is here, FAIL
___________________________________________________________________________________

I restored all the values in 'stdio'h and 'file.h' that I wrongly messed with while trying to increase _ftable from 10 to 15 entries, deleted rtsv7M4_T_le_v4SPD16_eabi.lib and 'cleaned' my project forcing rebuild but rename() still fails.

Do you have a list of 'most likely causes of rename() failure' I could investigate?
Thanks very much,
Dave
p.s.
I'm very happy with my system, as a speed test: able to plug 10 files into 10 virtual CPUs and have them all run 'concurrently' reading from 10 different files interpreting
for a = 0 to 10000
n = ((b x c) + (d + e) )SQU
next
at over 500 iterations per second on each cpu.

Hi Sean,

OK, got to single step thru ff.c...

In ff.c...

FRESULT f_rename (
        const TCHAR *path_old,  /* Pointer to the old name */
        const TCHAR *path_new   /* Pointer to the new name */
)
{
        FRESULT res;
        DIR djo, djn;
        BYTE buf[21], *dir;
        DWORD dw;
        DEF_NAMEBUF;


        res = chk_mounted(&path_old, &djo.fs, 1);
        if (res == FR_OK) {
                djn.fs = djo.fs;
                INIT_BUF(djo);
                res = follow_path(&djo, path_old);   /* Check old object */
                if (_FS_RPATH && res == FR_OK && (djo.fn[NS] & NS_DOT))
                        res = FR_INVALID_NAME;
#if _FS_SHARE
                if (res == FR_OK) res = chk_lock(&djo, 2);
#endif
                if (res == FR_OK) {                /* Old object is found */
                        if (!djo.dir) {            /* Is root dir? */
                                res = FR_NO_FILE;
                        } else {
                                mem_cpy(buf, djo.dir+DIR_Attr, 21); /* Save the object...
                                mem_cpy(&djn, &djo, sizeof(DIR));   /* Check new object */
                                res = follow_path(&djn, path_new); <----This line is returning 6 !

6 is:

Macro: int ENXIO

“No such device or address.” The system tried to use the device represented by a file you specified, and it couldn’t find the device. This can mean that the device file was installed incorrectly, or that the physical device is missing or not correctly attached to the computer".

Again, my function that calls rename():

// **************************************************************** TM4C123GH6PM
void renameFile(char *old, char *new){
char oldbuf[20];
char newbuf[20];

    sprintf(oldbuf, "fat:0:%s", old);
    sprintf(newbuf, "fat:0:%s", new);
    if(rename(oldbuf, newbuf) isNot 0){
        printB("RENAME FAILED \r%s to %s\n", oldbuf, newbuf);
        perror("RENAME");
//        return false;
    }
    else
        printB("Renamed \r%s to \r%s\n", oldbuf, newbuf);
//    return true;
}

Which passes "fat:0:TMP.BOB" and "fat:0:K.BOB" to rename().

Does this help?

I really don't want to stay with the work around.

Any ideas?

Cheers,

Dave

Hi Sean,
Would updating CCS wipe out my changes?
I'm worried my project (4 years in development) might be affected.
Is there a chance of that?
Thanks,
Dave