[LTP] [PATCH v5] coredump01: New core_pattern specifiers test

[Andrea Cervesato]

Hi Li,

> > +	TST_EXP_PASS(access(dump, F_OK), "core.%%e.%%p.%%s expanded to core.coredump01.%d.%d",
> > +		     pid, SIGABRT);
> 
> Generally this checks is really nice, but it'd be better to
> add more checks for the core dump file?
> 
> e.g. file size > 0, ELF magic valid, and type == ET_CORE.

this is already done, but we can introduce ELF type check (read below).

> 
> FYI:
> 
> +static int verify_core_dump(const char *path, long long expected_size)
> +{
> +       int fd;
> +       struct stat st;
> +       unsigned char magic[SELFMAG];
> +       Elf64_Ehdr ehdr;

Ehdr is not portable on 32 bit systems breaking our testing, so I
wouldn't use it in this case. We need Elf32_Half which is in both
32bit and 64bit.

> +
> +       if (access(path, F_OK) != 0) {
> +               tst_res(TFAIL, "Core dump file %s does not exist", path);
> +               return -1;
> +       }
> +       tst_res(TPASS, "Core dump file %s exists", path);
> +
> +       SAFE_STAT(path, &st);
> +       TST_EXP_EXPR(st.st_size > 0, "core file size = %ld bytes", st.st_size);
> +
> +       if (expected_size > 0 && st.st_size != expected_size) {
> +               tst_res(TFAIL, "Core dump size mismatch: expected %lld bytes, got %ld",
> +                       expected_size, st.st_size);
> +               return -1;
> +       }
> +
> +       fd = SAFE_OPEN(path, O_RDONLY, 0);
> +       SAFE_READ(0, fd, magic, SELFMAG);
> +       SAFE_CLOSE(fd);
> +
> +       TST_EXP_EXPR(!memcmp(magic, ELFMAG, 4),
> +                       "core file has valid ELF magic");
> +
> +       fd = SAFE_OPEN(path, O_RDONLY, 0);
> +       SAFE_READ(0, fd, &ehdr, sizeof(ehdr));
> +       SAFE_READ(0, fd, &ehdr, sizeof(ehdr));

double read? I guess it's an error

I will send a new version with some of the proposed checks

--
Andrea Cervesato
SUSE QE Automation Engineer Linux
andrea.cervesato@suse.com