8:37 a.m.
The "kmem -i" option may output the bogus statistics of CACHED, which
might be observed when some extreme situations occur in kernel, such as
OOM, disk IO errors, etc.
The following result of calculation may be a negative value, refer to
the dump_kmeminfo():
page_cache_size = nr_file_pages - swapper_space_nrpages - buffer_pages;
As a result, the negative value will be converted to unsigned long
integer, eventually it overflows and gets a big integer.
crash> kmem -i
PAGES TOTAL PERCENTAGE
TOTAL MEM 255314511 973.9 GB ----
FREE 533574 2 GB 0% of TOTAL MEM
USED 254780937 971.9 GB 99% of TOTAL MEM
SHARED 1713 6.7 MB 0% of TOTAL MEM
BUFFERS 374 1.5 MB 0% of TOTAL MEM
CACHED -114 70368744177664 GB 72251060080% of TOTAL MEM
^^^^ ^^^^^^^^^^^^^^ ^^^^^^^^^^^^
...
Let's normalize it to zero to fix such corner cases.
Reported-by: Buland Kumar Singh <bsingh(a)redhat.com>
Signed-off-by: Lianbo Jiang <lijiang(a)redhat.com>
---
memory.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/memory.c b/memory.c
index e0742c1bd3a4..860a3a978a4d 100644
--- a/memory.c
+++ b/memory.c
@@ -8615,6 +8615,8 @@ dump_kmeminfo(void)
page_cache_size = 0;
+ if (page_cache_size < 0)
+ page_cache_size = 0;
pct = (page_cache_size * 100)/totalram_pages;
fprintf(fp, "%13s %7ld %11s %3ld%% of TOTAL MEM\n",
"CACHED", page_cache_size,
--
2.37.1
11:38 p.m.
On 2023/02/14 23:37, Lianbo Jiang wrote:
The "kmem -i" option may output the bogus statistics of
CACHED, which
might be observed when some extreme situations occur in kernel, such as
OOM, disk IO errors, etc.
The following result of calculation may be a negative value, refer to
the dump_kmeminfo():
page_cache_size = nr_file_pages - swapper_space_nrpages - buffer_pages;
As a result, the negative value will be converted to unsigned long
integer, eventually it overflows and gets a big integer.
crash> kmem -i
PAGES TOTAL PERCENTAGE
TOTAL MEM 255314511 973.9 GB ----
FREE 533574 2 GB 0% of TOTAL MEM
USED 254780937 971.9 GB 99% of TOTAL MEM
SHARED 1713 6.7 MB 0% of TOTAL MEM
BUFFERS 374 1.5 MB 0% of TOTAL MEM
CACHED -114 70368744177664 GB 72251060080% of TOTAL MEM
^^^^ ^^^^^^^^^^^^^^ ^^^^^^^^^^^^
I think that this is not very bad output :) because we can notice
the negative value. Maybe it's due to future kernel changes.
So if you set it to zero, what about emitting an info message?
+ if (page_cache_size < 0) {
+ error(INFO, "page_cache_size went negative (%ld), setting to
0\n",
+ page_cache_size);
+ page_cache_size = 0;
+ }
For example,
crash> kmem -i
PAGES TOTAL PERCENTAGE
TOTAL MEM 16252500 62 GB ----
FREE 246162 961.6 MB 1% of TOTAL MEM
USED 16006338 61.1 GB 98% of TOTAL MEM
SHARED 329258 1.3 GB 2% of TOTAL MEM
BUFFERS 1436 5.6 MB 0% of TOTAL MEM
kmem: page_cache_size went negative (-114), setting to 0
CACHED 0 0 0% of TOTAL MEM
SLAB 311514 1.2 GB 1% of TOTAL MEM
Thanks,
Kazu
> ...
>
> Let's normalize it to zero to fix such corner cases.
>
> Reported-by: Buland Kumar Singh <bsingh(a)redhat.com>
> Signed-off-by: Lianbo Jiang <lijiang(a)redhat.com>
> ---
> memory.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/memory.c b/memory.c
> index e0742c1bd3a4..860a3a978a4d 100644
> --- a/memory.c
> +++ b/memory.c
> @@ -8615,6 +8615,8 @@ dump_kmeminfo(void)
> page_cache_size = 0;
>
>
> + if (page_cache_size < 0)
> + page_cache_size = 0;
> pct = (page_cache_size * 100)/totalram_pages;
> fprintf(fp, "%13s %7ld %11s %3ld%% of TOTAL MEM\n",
> "CACHED", page_cache_size,
12:35 a.m.
Hi, Kazu
Thank you for the comment.
On Wed, Feb 15, 2023 at 1:38 PM HAGIO KAZUHITO(萩尾 一仁) <k-hagio-ab(a)nec.com>
wrote:
It should be good to have an info message.
> + if (page_cache_size < 0) {
> + error(INFO, "page_cache_size went negative (%ld), setting
> to 0\n",
> + page_cache_size);
> + page_cache_size = 0;
> + }
Looks good.
Thanks.
Lianbo
On 2023/02/14 23:37, Lianbo Jiang wrote:
> The "kmem -i" option may output the bogus statistics of CACHED, which
> might be observed when some extreme situations occur in kernel, such as
> OOM, disk IO errors, etc.
>
> The following result of calculation may be a negative value, refer to
> the dump_kmeminfo():
> page_cache_size = nr_file_pages - swapper_space_nrpages -
buffer_pages;
>
> As a result, the negative value will be converted to unsigned long
> integer, eventually it overflows and gets a big integer.
>
> crash> kmem -i
> PAGES TOTAL PERCENTAGE
> TOTAL MEM 255314511 973.9 GB ----
> FREE 533574 2 GB 0% of TOTAL MEM
> USED 254780937 971.9 GB 99% of TOTAL MEM
> SHARED 1713 6.7 MB 0% of TOTAL MEM
> BUFFERS 374 1.5 MB 0% of TOTAL MEM
> CACHED -114 70368744177664 GB 72251060080% of TOTAL MEM
> ^^^^ ^^^^^^^^^^^^^^ ^^^^^^^^^^^^
I think that this is not very bad output :) because we can notice
the negative value. Maybe it's due to future kernel changes.
It's true. But it is also easy to mislead users, because it is not clear
whether the crash tool
gave a wrong result.
Actually this issue was observed on the kernel 3.10.
So if you set it to zero, what about emitting an info message?
For example,
crash> kmem -i
PAGES TOTAL PERCENTAGE
TOTAL MEM 16252500 62 GB ----
FREE 246162 961.6 MB 1% of TOTAL MEM
USED 16006338 61.1 GB 98% of TOTAL MEM
SHARED 329258 1.3 GB 2% of TOTAL MEM
BUFFERS 1436 5.6 MB 0% of TOTAL MEM
kmem: page_cache_size went negative (-114), setting to 0
CACHED 0 0 0% of TOTAL MEM
SLAB 311514 1.2 GB 1% of TOTAL MEM
Thanks,
Kazu
> ...
>
> Let's normalize it to zero to fix such corner cases.
>
> Reported-by: Buland Kumar Singh <bsingh(a)redhat.com>
> Signed-off-by: Lianbo Jiang <lijiang(a)redhat.com>
> ---
> memory.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/memory.c b/memory.c
> index e0742c1bd3a4..860a3a978a4d 100644
> --- a/memory.c
> +++ b/memory.c
> @@ -8615,6 +8615,8 @@ dump_kmeminfo(void)
> page_cache_size = 0;
>
>
> + if (page_cache_size < 0)
> + page_cache_size = 0;
> pct = (page_cache_size * 100)/totalram_pages;
> fprintf(fp, "%13s %7ld %11s %3ld%% of TOTAL MEM\n",
> "CACHED", page_cache_size,
1:07 a.m.
On 2023/02/15 15:35, lijiang wrote:>>> Hi, Kazu
>> Thank you for the comment.
>>
>> On Wed, Feb 15, 2023 at 1:38 PM HAGIO KAZUHITO(萩尾 一仁) <k-hagio-ab(a)nec.com
<mailto:k-hagio-ab@nec.com>> wrote:
>>
>> On 2023/02/14 23:37, Lianbo Jiang wrote:
>> > The "kmem -i" option may output the bogus statistics of
CACHED, which
>> > might be observed when some extreme situations occur in kernel, such
as
>> > OOM, disk IO errors, etc.
>> >
>> > The following result of calculation may be a negative value, refer to
>> > the dump_kmeminfo():
>> > page_cache_size = nr_file_pages - swapper_space_nrpages -
buffer_pages;
>> >
>> > As a result, the negative value will be converted to unsigned long
>> > integer, eventually it overflows and gets a big integer.
>> >
>> > crash> kmem -i
>> > PAGES TOTAL PERCENTAGE
>> > TOTAL MEM 255314511 973.9 GB ----
>> > FREE 533574 2 GB 0% of TOTAL MEM
>> > USED 254780937 971.9 GB 99% of TOTAL MEM
>> > SHARED 1713 6.7 MB 0% of TOTAL MEM
>> > BUFFERS 374 1.5 MB 0% of TOTAL MEM
>> > CACHED -114 70368744177664 GB 72251060080% of TOTAL
MEM
>> > ^^^^ ^^^^^^^^^^^^^^ ^^^^^^^^^^^^
>>
>> I think that this is not very bad output :) because we can notice
>> the negative value. Maybe it's due to future kernel changes.
>>
>>
>> It's true. But it is also easy to mislead users, because it is not clear
whether the crash tool
>> gave a wrong result.
>>
>> Actually this issue was observed on the kernel 3.10.
>>
>> So if you set it to zero, what about emitting an info message?
>>
>>
>> It should be good to have an info message.
>>
>> + if (page_cache_size < 0) {
>> + error(INFO, "page_cache_size went negative (%ld),
setting to 0\n",
>> + page_cache_size);
>> + page_cache_size = 0;
>> + }
>>
>>
>> Looks good.
ok thanks for checking, applied.
https://github.com/crash-utility/crash/commit/277da34dd5da8c1280d0d0fd7ce...
Thanks,
Kazu
681
days inactive
682
days old
devel@lists.crash-utility.osci.io
3 comments
3 participants
participants (3)
-
HAGIO KAZUHITO(萩尾 一仁) -
Lianbo Jiang -
lijiang