phoenix-rtos / phoenix-rtos-utils Goto Github PK

phoenix-rtos-utils/core/psh/mem/mem.c: In function 'psh_mem_procprint':
phoenix-rtos-utils/core/psh/mem/mem.c:85:20: warning: format '%llu' expects argument of type 'long long unsigned int', but argument 3 has type 'id_t' {aka 'unsigned int'} [-Wformat=]
   85 |    printf("  %d.%llu", e->oid.port, e->oid.id);
      |                 ~~~^                ~~~~~~~~~
      |                    |                      |
      |                    long long unsigned int id_t {aka unsigned int}
      |                 %u

Example build from ghactions: https://github.com/phoenix-rtos/phoenix-rtos-project/runs/2792844210?check_suite_focus=true#step:4:260

Output is incorrectly formatted when using some commands on 64-bit architectures (e.g. mem -m on riscv64). Screenshot included.

Variable op_seecondary is always set to 1.

Target: ia32-generic

Hangs on second execution from ash.

@niewim19 can You check if it's connected with #44 or if it's a separate issue?

Maybe we could add quick reproduction test for this case before fixing it?

At this moment, when trying to execute a sysexec command with a syspage program, which is not on the white list we print a message Unknown command. It is misleading, as it suggests to the user, that the command itself does not exist or he/she made a mistake when executing the command. We should change it to for example sysexec: Command not allowed or even explicitly: sysexec: Program micropython not found on the whitelist, whichever you prefer.

Sometimes (approx 20%) ERASE_ALL falsifely detects bad blocks (race condition / cache issue?):

WRITE_REGISTER -3 128 8

Currently psh exits always with 0 which makes any of the applets unusable in complex environment.

Also: revisit psh applet return codes, as only the lower 8 bits are passed to the parent process - so returning negative values might not be the best idea.

For details see: https://pubs.opengroup.org/onlinepubs/007904875/functions/exit.html

Move flashsrv dependent code to separate directories (e.g. nand, nor)

psh/psh.c:

int main(int argc, char **argv)
...
		if ((args = calloc(3000, 1)) == NULL) {  
			printf("psh: out of memory\n");  
			return EXIT_FAILURE;  
		}  
  
		for (c = 1; c < argc; ++c) {  
			strcat(args, argv[c]);  
			strcat(args, " ");  
		}
...

Appending to args is done without any control of the possible args buffer overflow.
What happens if the sum of the length of the arguments is greater than 3000?
Why the number 3000? Can it be greater than 3000 at all?
In psh_run is 128

void psh_run(void)
...
	char buf[128];

It would probably be appropriate to use the :

strncat(char *, const char *, size_t)

but now this function does nothing.
libphoenix/string/string.c

char *strncat(char *s1, const char *s2i, size_t num)  
{  
	return 0;  
}

I also don't see:

free(args);

When we have hundrets of ms wait time, ps outputs it as eg. 383.s instead of 383ms which looks really worrisome.

Expected output for ls -d:

But got: