-----Original Message-----
From: Richard W.M. Jones [mailto:rjones@redhat.com]
Sent: Monday, March 23, 2015 9:29 PM
To: Chen, Hanxiao/陈 晗霄
Cc: libguestfs(a)redhat.com
Subject: Re: [Libguestfs] [PATCH] New API: part_get_part_type for showing partition
type
On Tue, Mar 17, 2015 at 02:45:46AM -0400, Chen Hanxiao wrote:
> This patch will add support for getting partition type
> of a partiton numbered device.
>
> Signed-off-by: Chen Hanxiao <chenhanxiao(a)cn.fujitsu.com>
> ---
> daemon/parted.c | 112
+++++++++++++++++++++++++++++++++++++++++++++++++++
> generator/actions.ml | 18 +++++++++
> src/MAX_PROC_NR | 2 +-
> 3 files changed, 131 insertions(+), 1 deletion(-)
>
> diff --git a/daemon/parted.c b/daemon/parted.c
> index a7bcb99..0ae6e5c 100644
> --- a/daemon/parted.c
> +++ b/daemon/parted.c
> @@ -33,6 +33,10 @@ GUESTFSD_EXT_CMD(str_parted, parted);
> GUESTFSD_EXT_CMD(str_sfdisk, sfdisk);
> GUESTFSD_EXT_CMD(str_sgdisk, sgdisk);
>
> +#ifndef PARTED_NO_M
> +# define PARTED_NO_M 0
> +#endif
What does this bit do?
I want to use print_partition_table(xxx, 0)
but do not want to pass a raw '0'.
Maybe the name is not so clear.
I'll rename it to PARTED_NOT_USE_M in the next version.
> /* Notes:
> *
> * Parted 1.9 sends error messages to stdout, hence use of the
> @@ -1022,3 +1026,111 @@ do_part_get_name (const char *device, int partnum)
> reply_with_error ("cannot get the partition name from '%s'
layouts", parttype);
> return NULL;
> }
> +
> +char *
> +do_part_get_part_type (const char *device, int partnum)
> +{
> + CLEANUP_FREE char *parttype;
> + char *part_type;
> +
> + if (partnum <= 0) {
> + reply_with_error ("partition number must be >= 1");
> + return NULL;
> + }
> +
> + parttype = do_part_get_parttype (device);
> + if (parttype == NULL)
> + return NULL;
> +
> + if (STREQ (parttype, "gpt")) {
> + part_type = strdup("primary");
> + if (part_type == NULL) {
> + reply_with_error ("strdup failed");
> + return NULL;
> + }
> + return part_type;
> + }
> +
> + /* machine parseable output by 'parted -m' did not provide
> + * partition type info.
> + * Use traditional style.
> + */
> + CLEANUP_FREE char *out = print_partition_table (device, PARTED_NO_M);
> + if (!out)
> + return NULL;
> +
> + CLEANUP_FREE_STRING_LIST char **lines = split_lines (out);
> +
> + if (!lines)
> + return NULL;
> +
> + size_t start = 0, end = 0, row;
> +
> + for (row = 0; lines[row] != NULL; ++row)
> + if (STRPREFIX (lines[row], "Number")) {
> + start = row + 1;
> + break;
> + }
> +
> + if (start == 0) {
> + reply_with_error ("parted output has no \"Number\" line");
> + return NULL;
> + }
> +
> + for (row = start; lines[row] != NULL; ++row)
> + if (STREQ (lines[row], "")) {
> + end = row;
> + break;
> + }
> +
> + if (end == 0) {
> + reply_with_error ("parted output has no blank after end of table");
> + return NULL;
> + }
> +
> + /* Now parse the lines. */
> + size_t i;
> + int64_t temp_int64;
> + int part_num;
> + char temp_type[16];
> + for (i = 0, row = start; row < end; ++i, ++row) {
> + if (sscanf (lines[row], "%d%" SCNi64 "B%" SCNi64
"B%" SCNi64 "B" "%s",
> + &part_num,
> + &temp_int64,
> + &temp_int64,
> + &temp_int64,
> + temp_type) != 5) {
We're hoping here that the output of 'parted' never changes and
overflows the 16 byte buffer.
Instead of hoping, you can guarantee that by replacing %s with %15s
Will fix.
Regards,
- Chen
> + reply_with_error ("could not parse row from output of parted print
command: %s", lines[row]);
> + return NULL;
> + }
> +
> + if (part_num != partnum)
> + continue;
> +
> + if (STRPREFIX (temp_type, "primary")) {
> + part_type = strdup("primary");
> + if (part_type == NULL)
> + goto error;
> + } else if (STRPREFIX (temp_type, "logical")) {
> + part_type = strdup("logical");
> + if (part_type == NULL)
> + goto error;
> + } else if (STRPREFIX (temp_type, "extended")) {
> + part_type = strdup("extended");
> + if (part_type == NULL)
> + goto error;
> + } else
> + goto error;
> +
> + return part_type;
> + }
> +
> + if (row == end) {
> + reply_with_error ("could not find partnum: %d", partnum);
> + return NULL;
> + }
> +
> + error:
> + reply_with_error ("strdup failed");
> + return NULL;
> +}
> diff --git a/generator/actions.ml b/generator/actions.ml
> index fb971d3..72418b0 100644
> --- a/generator/actions.ml
> +++ b/generator/actions.ml
> @@ -12522,6 +12522,24 @@ This will Enable extended inode refs." };
> longdesc = "\
> This enable skinny metadata extent refs." };
>
> + { defaults with
> + name = "part_get_part_type";
> + style = RString "partitiontype", [Device "device"; Int
"partnum"], [];
> + proc_nr = Some 453;
> + tests = [
> + InitEmpty, Always, TestResultString (
> + [["part_init"; "/dev/sda"; "mbr"];
> + ["part_add"; "/dev/sda"; "p";
"64"; "204799"];
> + ["part_add"; "/dev/sda"; "e";
"204800"; "614400"];
> + ["part_add"; "/dev/sda"; "l";
"204864"; "205988"];
> + ["part_get_part_type"; "/dev/sda"; "5"]],
"logical"), []
> + ];
> +
> + shortdesc = "get the partition type";
> + longdesc = "\
> +This could get the partition type, such as primary, logical,
> +on partition numbered C<partnum> on device C<device>." };
> +
> ]
>
> (* Non-API meta-commands available only in guestfish.
> diff --git a/src/MAX_PROC_NR b/src/MAX_PROC_NR
> index 8670c73..534b992 100644
> --- a/src/MAX_PROC_NR
> +++ b/src/MAX_PROC_NR
> @@ -1 +1 @@
> -452
> +453
Rich.
--
Richard Jones, Virtualization Group, Red Hat
http://people.redhat.com/~rjones
Read my programming and virtualization blog:
http://rwmj.wordpress.com
virt-top is 'top' for virtual machines. Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top