[PATCH] page-types: decode flags directly from command line

[Alex Chiang]

Teach page-types to decode page flags directly from the command
line.

Why is this useful? For instance, if you're using memory hotplug
and see this in /var/log/messages:

kernel: removing from LRU failed 3836dd0/1/1e00000000000400

It would be nice to decode those page flags without staring at
the source.

Example usage and output:

linux-2.6/Documentation/vm$ ./page-types -d 0x1e00000000000400
flags page-count MB symbolic-flags long-symbolic-flags
0x1e00000000000400 1 0 __________B_______________________buddy
total 1 0

Signed-off-by: Alex Chiang <ach...@hp.com>
---
page-types.c | 26 ++++++++++++++++++++++++--
1 file changed, 24 insertions(+), 2 deletions(-)
---
diff --git a/Documentation/vm/page-types.c b/Documentation/vm/page-types.c
index 3ec4f2a..a55c624 100644
--- a/Documentation/vm/page-types.c
+++ b/Documentation/vm/page-types.c
@@ -674,6 +674,7 @@ static void usage(void)
printf(
"page-types [options]\n"
" -r|--raw Raw mode, for kernel developers\n"
+" -d|--decode flags Decode a single page's flags\n"
" -a|--addr addr-spec Walk a range of pages\n"
" -b|--bits bits-spec Walk pages with specified bits\n"
" -p|--pid pid Walk process address space\n"
@@ -682,10 +683,12 @@ static void usage(void)
#endif
" -l|--list Show page details in ranges\n"
" -L|--list-each Show page details one by one\n"
-" -N|--no-summary Don't show summay info\n"
+" -N|--no-summary Don't show summary info\n"
" -X|--hwpoison hwpoison pages\n"
" -x|--unpoison unpoison pages\n"
" -h|--help Show this usage message\n"
+"flags:\n"
+" 0x0000000000000400 A single page's flags, e.g.\n"
"addr-spec:\n"
" N one page at offset N (unit: pages)\n"
" N+M pages range from N to N+M-1\n"
@@ -884,12 +887,28 @@ static void parse_bits_mask(const char *optarg)
add_bits_filter(mask, bits);
}

+static void decode_flags_and_exit(const char *optarg)
+{
+ uint64_t flags;
+
+ flags = parse_number(optarg);
+
+ opt_list = 0;
+ opt_hwpoison = 0;
+ opt_unpoison = 0;
+
+ add_page(0, 0, flags);
+ show_summary();
+
+ exit(0);
+}

static struct option opts[] = {
{ "raw" , 0, NULL, 'r' },
{ "pid" , 1, NULL, 'p' },
{ "file" , 1, NULL, 'f' },
{ "addr" , 1, NULL, 'a' },
+ { "decode" , 1, NULL, 'd' },
{ "bits" , 1, NULL, 'b' },
{ "list" , 0, NULL, 'l' },
{ "list-each" , 0, NULL, 'L' },
@@ -907,7 +926,7 @@ int main(int argc, char *argv[])
page_size = getpagesize();

while ((c = getopt_long(argc, argv,
- "rp:f:a:b:lLNXxh", opts, NULL)) != -1) {
+ "rp:f:a:b:d:lLNXxh", opts, NULL)) != -1) {
switch (c) {
case 'r':
opt_raw = 1;
@@ -924,6 +943,9 @@ int main(int argc, char *argv[])
case 'b':
parse_bits_mask(optarg);
break;
+ case 'd':
+ decode_flags_and_exit(optarg);
+ break;
case 'l':
opt_list = 1;
break;
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majo...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

[Wu Fengguang]

Hi Alex,

On Wed, Nov 04, 2009 at 06:54:41AM +0800, Alex Chiang wrote:
> Teach page-types to decode page flags directly from the command
> line.

This is good feature to have - I considered adding it, too ;)

>
> Why is this useful? For instance, if you're using memory hotplug
> and see this in /var/log/messages:
>
> kernel: removing from LRU failed 3836dd0/1/1e00000000000400
>
> It would be nice to decode those page flags without staring at
> the source.

In fact it's more than decode - encoding is also possible with the
_same_ code! So maybe "-d" and help message will not be all that
appropriate.

> Example usage and output:
>
> linux-2.6/Documentation/vm$ ./page-types -d 0x1e00000000000400
> flags page-count MB symbolic-flags long-symbolic-flags
> 0x1e00000000000400 1 0 __________B_______________________buddy
> total 1 0

The output is a bit redundant - so does the code. Could you simplify
them a bit?

Thanks,
Fengguang

[Wu Fengguang]

Hi Alex,

It's easier to illustrate the idea with example and code :)

# Documentation/vm/page-types -d 0x10,anon
0x0000000000001010 ____D_______a_____________________ dirty,anonymous

+static void describe_flags(const char *optarg)
+{
+ uint64_t flags = parse_flag_names(optarg, 0);
+
+ printf("0x%016llx\t%s\t%s\n",
+ (unsigned long long)flags,
+ page_flag_name(flags),
+ page_flag_longname(flags));
+}

+ case 'd':
+ opt_no_summary = 1;
+ describe_flags(optarg);
+ break;

Thanks,
Fengguang

On Thu, Nov 05, 2009 at 04:40:08AM +0800, Alex Chiang wrote:
> Hi Fengguang,
>
> * Wu Fengguang <fenggu...@intel.com>:

> > On Wed, Nov 04, 2009 at 06:54:41AM +0800, Alex Chiang wrote:
> > > Why is this useful? For instance, if you're using memory hotplug
> > > and see this in /var/log/messages:
> > >
> > > kernel: removing from LRU failed 3836dd0/1/1e00000000000400
> > >
> > > It would be nice to decode those page flags without staring at
> > > the source.
> >
> > In fact it's more than decode - encoding is also possible with the
> > _same_ code! So maybe "-d" and help message will not be all that
> > appropriate.
>

> I'm sorry, I don't understand this use case, so I'm not sure what
> you're asking me to do.
>
> You're saying that a use case would be something like:
>
> ./page-types --encode referenced,mmap
> 0x0000000000000004
>
> ?
>
> If that's what you're asking for, I guess I'm not sure why that's
> so useful, but then again, I'm a vm n00b so there are probably
> lots of things I don't understand. ;)

>
> > > Example usage and output:
> > >
> > > linux-2.6/Documentation/vm$ ./page-types -d 0x1e00000000000400
> > > flags page-count MB symbolic-flags long-symbolic-flags
> > > 0x1e00000000000400 1 0 __________B_______________________buddy
> > > total 1 0
> >
> > The output is a bit redundant - so does the code. Could you simplify
> > them a bit?
>

> Well, the code is redundant, but add_page() / show_summary() is a
> simple sequence.
>
> In contrast, I think I'd have to modify walk_addr_ranges() and
> maybe walk_pfn() to do something special when we don't really
> want to do any address space walking, and simply want to
> decode/encode some user input.
>
> Maybe I don't understand you fully? Could you give me a better
> idea of what you're looking for?
>
> As for the output, I'm just reusing show_summary(). Maybe we
> don't need the flags, page-count, and MB columns, but again, the
> patch would be more intrusisive because we'd have to teach
> show_summary() about the special case.
>
> Anyway, I'm happy to make changes closer to what you're looking
> for, but I'd like some more guidance as to what you're expecting.
>
> Thanks,
> /ac

[Alex Chiang]

Hi Fengguang,

* Wu Fengguang <fenggu...@intel.com>:

> On Wed, Nov 04, 2009 at 06:54:41AM +0800, Alex Chiang wrote:
> > Why is this useful? For instance, if you're using memory hotplug
> > and see this in /var/log/messages:
> >
> > kernel: removing from LRU failed 3836dd0/1/1e00000000000400
> >
> > It would be nice to decode those page flags without staring at
> > the source.
>
> In fact it's more than decode - encoding is also possible with the
> _same_ code! So maybe "-d" and help message will not be all that
> appropriate.

I'm sorry, I don't understand this use case, so I'm not sure what

you're asking me to do.

You're saying that a use case would be something like:

./page-types --encode referenced,mmap
0x0000000000000004

?

If that's what you're asking for, I guess I'm not sure why that's
so useful, but then again, I'm a vm n00b so there are probably
lots of things I don't understand. ;)

> > Example usage and output:

> >
> > linux-2.6/Documentation/vm$ ./page-types -d 0x1e00000000000400
> > flags page-count MB symbolic-flags long-symbolic-flags
> > 0x1e00000000000400 1 0 __________B_______________________buddy
> > total 1 0
>
> The output is a bit redundant - so does the code. Could you simplify
> them a bit?

Well, the code is redundant, but add_page() / show_summary() is a
simple sequence.

In contrast, I think I'd have to modify walk_addr_ranges() and
maybe walk_pfn() to do something special when we don't really
want to do any address space walking, and simply want to
decode/encode some user input.

Maybe I don't understand you fully? Could you give me a better
idea of what you're looking for?

As for the output, I'm just reusing show_summary(). Maybe we
don't need the flags, page-count, and MB columns, but again, the
patch would be more intrusisive because we'd have to teach
show_summary() about the special case.

Anyway, I'm happy to make changes closer to what you're looking
for, but I'd like some more guidance as to what you're expecting.

Thanks,
/ac

--