Re: [Qemu-block] [PATCH v7] qemu-io: add pattern file for write command

[Max Reitz]

On 05.07.19 12:21, Denis Plotnikov wrote:
> The patch allows to provide a pattern file for write
> command. There was no similar ability before.
> 
> Signed-off-by: Denis Plotnikov <address@hidden>
> ---
> v7:
>   * fix variable naming
>   * make code more readable
>   * extend help for write command
> 
> v6:
>   * the pattern file is read once to reduce io
> 
> v5:
>   * file name initiated with null to make compilers happy
> 
> v4:
>   * missing signed-off clause added
> 
> v3:
>   * missing file closing added
>   * exclusive flags processing changed
>   * buffer void* converted to char* to fix pointer arithmetics
>   * file reading error processing added
> ---
>  qemu-io-cmds.c | 86 ++++++++++++++++++++++++++++++++++++++++++++++----
>  1 file changed, 80 insertions(+), 6 deletions(-)
> 
> diff --git a/qemu-io-cmds.c b/qemu-io-cmds.c
> index 09750a23ce..495170380a 100644
> --- a/qemu-io-cmds.c
> +++ b/qemu-io-cmds.c
> @@ -343,6 +343,66 @@ static void *qemu_io_alloc(BlockBackend *blk, size_t 
> len, int pattern)
>      return buf;
>  }
>  
> +static void *qemu_io_alloc_from_file(BlockBackend *blk, size_t len,
> +                                     char *file_name)

Wouldn’t a const char * suffice?

> +{
> +    char *buf, *buf_origin;
> +    FILE *f = fopen(file_name, "r");
> +    int pattern_len;
> +
> +    if (!f) {
> +        printf("'%s': %s\n", file_name, strerror(errno));

(1) Why stdout and not stderr?

(2) Why not just perror()?

> +        return NULL;
> +    }
> +
> +    if (qemuio_misalign) {
> +        len += MISALIGN_OFFSET;
> +    }
> +
> +    buf_origin = buf = blk_blockalign(blk, len);
> +
> +    pattern_len = fread(buf, sizeof(char), len, f);

I know I pointed out already that the C standard defines sizeof(char) to
be 1, but I just want to add that “sizeof(char)” does not appear
anywhere in qemu so far.

That doesn’t make it wrong, but here it looks a bit weird...

> +
> +    if (ferror(f)) {
> +        printf("'%s': %s\n", file_name, strerror(errno));
> +        goto error;
> +    }
> +
> +    if (pattern_len == 0) {
> +        printf("'%s' is empty\n", file_name);
> +        goto error;
> +    }
> +
> +    fclose(f);
> +
> +    if (len > pattern_len) {
> +        char *file_buf = g_malloc(sizeof(char) * pattern_len);

...and here you’d actually have to check that the multiplication doesn’t
overflow if you actually consider it a possibility that
sizeof(char) != 1.

But actually, this allocation is just unnecessary altogether, as
Vladimir has said for v6.  You can drop this line,...

> +        memcpy(file_buf, buf, pattern_len);

...this line,...

> +        len -= pattern_len;
> +        buf += pattern_len;
> +
> +        while (len > 0) {
> +            size_t len_to_copy = MIN(pattern_len, len);
> +
> +            memcpy(buf, file_buf, len_to_copy);

...make this a memcpy(buf, buf_origin, len_to_copy);,...

> +
> +            len -= len_to_copy;
> +            buf += len_to_copy;
> +        }
> +        qemu_vfree(file_buf);

...and drop this line, too.

> +    }
> +
> +    if (qemuio_misalign) {
> +        buf_origin += MISALIGN_OFFSET;

This will change the pattern, though, because now you’re skipping the
first MISALIGN_OFFSET bytes.  Should the fread() above read to
buf_origin + MISALIGN_OFFSET in that case?

[...]

> @@ -983,8 +1044,9 @@ static int write_f(BlockBackend *blk, int argc, char 
> **argv)
>      /* Some compilers get confused and warn if this is not initialized.  */
>      int64_t total = 0;
>      int pattern = 0xcd;
> +    char *file_name = NULL;
>  
> -    while ((c = getopt(argc, argv, "bcCfnpP:quz")) != -1) {
> +    while ((c = getopt(argc, argv, "bcCfnpP:quzs:")) != -1) {
>          switch (c) {
>          case 'b':
>              bflag = true;
> @@ -1020,6 +1082,10 @@ static int write_f(BlockBackend *blk, int argc, char 
> **argv)
>          case 'z':
>              zflag = true;
>              break;
> +        case 's':
> +            sflag = true;
> +            file_name = g_strdup(optarg);

Why does file_name need to be a char * and duplicated here?  (optarg
always points into argv, so as long as that doesn’t change, the optarg
pointer stays valid.)

Max

> +            break;
>          default:
>              qemuio_command_usage(&write_cmd);
>              return -EINVAL;

signature.asc
Description: OpenPGP digital signature