[PATCH] io/mremap: modify argument errors of mremap command
Brian Foster
bfoster at redhat.com
Wed Mar 16 07:31:26 CDT 2016
On Wed, Mar 16, 2016 at 05:31:47PM +0800, Zorro Lang wrote:
> There're 2 argument errors in mremap command:
> 1. If run "mremap 1024 8192", it won't return error and try to
> do "mremap 1024" silently.
>
> 2. The "-f" option can't be used, due to it need a new address
> argument as the fifth argument of mremap() syscall(man mremap).
>
> This patch try to fix above two problems.
>
> Signed-off-by: Zorro Lang <zlang at redhat.com>
> ---
A couple nits...
>
> Hi,
>
> I'm not sure pass a address argument to mremap command is right or
> wrong. And maybe I should use strtoul() to instead of cvtnum() for
> address argument.
>
> At least I did some simple test with this patch on my machine, results are:
> "mremap -f 139946004389888 2048" PASS
> "mremap -f 4096k 2048" PASS
> "mremap -f 1g 2048" PASS
> "mremap -f 4096 2048" FAIL as 'Permission denied'
>
> What do you think of this patch?
>
> Thanks,
> Zorro
>
>
> io/mmap.c | 22 +++++++++++++++-------
> man/man8/xfs_io.8 | 5 ++++-
> 2 files changed, 19 insertions(+), 8 deletions(-)
>
> diff --git a/io/mmap.c b/io/mmap.c
> index 5970069..672ee61 100644
> --- a/io/mmap.c
> +++ b/io/mmap.c
> @@ -589,7 +589,7 @@ mremap_help(void)
> "\n"
> " Resizes the mappping, growing or shrinking from the current size.\n"
> " The default stored value is 'X', repeated to fill the range specified.\n"
> -" -f -- use the MREMAP_FIXED flag\n"
> +" -f <new_address> -- use the MREMAP_FIXED flag to mremap on the new address\n"
Exceeds 80 chars.
> " -m -- use the MREMAP_MAYMOVE flag\n"
> "\n"));
> }
> @@ -600,15 +600,18 @@ mremap_f(
> char **argv)
> {
> ssize_t new_length;
> - void *new_addr;
> + void *new_addr = NULL;
> int flags = 0;
> int c;
> size_t blocksize, sectsize;
>
> - while ((c = getopt(argc, argv, "fm")) != EOF) {
> + init_cvtnum(&blocksize, §size);
> +
> + while ((c = getopt(argc, argv, "f:m")) != EOF) {
> switch (c) {
> case 'f':
> flags = MREMAP_FIXED|MREMAP_MAYMOVE;
> + new_addr = (void *)cvtnum(blocksize, sectsize, optarg);
> break;
> case 'm':
> flags = MREMAP_MAYMOVE;
> @@ -618,7 +621,9 @@ mremap_f(
> }
> }
>
> - init_cvtnum(&blocksize, §size);
> + if (optind != argc - 1)
> + return command_usage(&mremap_cmd);
> +
> new_length = cvtnum(blocksize, sectsize, argv[optind]);
> if (new_length < 0) {
> printf(_("non-numeric offset argument -- %s\n"),
> @@ -626,7 +631,10 @@ mremap_f(
> return 0;
> }
>
> - new_addr = mremap(mapping->addr, mapping->length, new_length, flags);
> + if (!new_addr)
> + new_addr = mremap(mapping->addr, mapping->length, new_length, flags);
> + else
> + new_addr = mremap(mapping->addr, mapping->length, new_length, flags, new_addr);
Both mremap() calls above exceed 80 chars. Otherwise, this looks Ok to
me:
Reviewed-by: Brian Foster <bfoster at redhat.com>
> if (new_addr == MAP_FAILED)
> perror("mremap");
> else {
> @@ -697,9 +705,9 @@ mmap_init(void)
> mremap_cmd.altname = "mrm";
> mremap_cmd.cfunc = mremap_f;
> mremap_cmd.argmin = 1;
> - mremap_cmd.argmax = 2;
> + mremap_cmd.argmax = 3;
> mremap_cmd.flags = CMD_NOFILE_OK | CMD_FOREIGN_OK;
> - mremap_cmd.args = _("[-m|-f] newsize");
> + mremap_cmd.args = _("[-m|-f <new_address>] newsize");
> mremap_cmd.oneline =
> _("alters the size of the current memory mapping");
> mremap_cmd.help = mremap_help;
> diff --git a/man/man8/xfs_io.8 b/man/man8/xfs_io.8
> index 33fbe6a..984b551 100644
> --- a/man/man8/xfs_io.8
> +++ b/man/man8/xfs_io.8
> @@ -581,7 +581,7 @@ See the
> .B mmap
> command.
> .TP
> -.BI "mremap [ \-f ] [ \-m ] " new_length
> +.BI "mremap [ \-f <new_address> ] [ \-m ] " new_length
> Changes the current mapping size to
> .IR new_length .
> Whether the mapping may be moved is controlled by the flags passed;
> @@ -589,6 +589,9 @@ MREMAP_FIXED
> .RB ( \-f ),
> or MREMAP_MAYMOVE
> .RB ( \-m ).
> +.IR new_length
> +specifies a page-aligned address to which the mapping must be moved. It
> +can be setted to 139946004389888, 4096k or 1g etc.
> .TP
> .B mrm
> See the
> --
> 2.5.0
>
> _______________________________________________
> xfs mailing list
> xfs at oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
More information about the xfs
mailing list