From: starlight@binnacle.cx
To: Mel Gorman <mel@csn.ul.ie>
Cc: Andrew Morton <akpm@linux-foundation.org>,
linux-mm@kvack.org, bugzilla-daemon@bugzilla.kernel.org,
bugme-daemon@bugzilla.kernel.org, Adam Litke <agl@us.ibm.com>,
Eric B Munson <ebmunson@us.ibm.com>
Subject: Re: [Bugme-new] [Bug 13302] New: "bad pmd" on fork() of process with hugepage shared memory segments attached
Date: Fri, 15 May 2009 14:44:29 -0400 [thread overview]
Message-ID: <6.2.5.6.2.20090515144058.03a55298@binnacle.cx> (raw)
[-- Attachment #1: Type: text/plain, Size: 476 bytes --]
This was really bugging me, so I hacked out
the test case for the attach failure.
Hoses 2.6.29.1 100% every time. Run it like this:
tcbm_att
tcbm_att -
tcbm_att -
tcbm_att -
It will break on the last iteration with ENOMEM
and ENOMEM is all any shmget() or shmat() call
gets forever more.
After removing the segments this appears:
HugePages_Total: 2048
HugePages_Free: 2048
HugePages_Rsvd: 1280
HugePages_Surp: 0
Even though no segments show in 'ipcs'.
[-- Attachment #2: tcbm_att.C.txt --]
[-- Type: text/plain, Size: 2429 bytes --]
extern "C" {
#include <errno.h>
#include <memory.h>
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>
#include <sys/shm.h>
}
int main(
int argc,
const char** argv,
const char** envp
)
{
if (argc == 1) {
int seg1id = shmget(0x12345600,
(size_t) 0x40000000,
IPC_CREAT|SHM_HUGETLB|0640
);
if (seg1id == -1) {
perror("shmget(1GB)");
return 1;
}
void* seg1adr = shmat(seg1id, (void*) 0x400000000, 0);
if (seg1adr == (void*) -1) {
perror("shmat(1GB)");
return 1;
}
int seg2id = shmget(0x12345601,
(size_t) 0x10000000,
IPC_CREAT|SHM_HUGETLB|0640
);
if (seg2id == -1) {
perror("shmget(256MB)");
return 1;
}
void* seg2adr = shmat(seg2id, (void*) 0x580000000, 0);
if (seg2adr == (void*) -1) {
perror("shmat(256MB)");
return 1;
}
char* seg_p = (char*) seg1adr;
int i1 = 182;
while (i1 > 0) {
memset(seg_p, 0x55, 0x400000);
seg_p += 0x400000;
i1--;
}
seg_p = (char*) seg2adr;
i1 = 6;
while (i1 > 0) {
memset(seg_p, 0xAA, 0x400000);
seg_p += 0x400000;
i1--;
}
if (shmdt((void*) 0x400000000) != 0) {
perror("shmdt(1GB)");
return 1;
}
if (shmdt((void*) 0x580000000) != 0) {
perror("shmdt(256MB)");
return 1;
}
} else {
int seg1id = shmget(0x12345600, 0, 0);
if (seg1id == -1) {
perror("shmget(1GB)");
return 1;
}
void* seg1adr = shmat(seg1id, (void*) 0x400000000, SHM_RDONLY);
if (seg1adr == (void*) -1) {
perror("shmat(1GB)");
return 1;
}
int seg2id = shmget(0x12345601, 0, 0);
if (seg2id == -1) {
perror("shmget(256MB)");
return 1;
}
void* seg2adr = shmat(seg2id, (void*) 0x580000000, SHM_RDONLY);
if (seg2adr == (void*) -1) {
perror("shmat(256MB)");
return 1;
}
if (shmdt((void*) 0x400000000) != 0) {
perror("shmdt(1GB)");
return 1;
}
if (shmdt((void*) 0x580000000) != 0) {
perror("shmdt(256MB)");
return 1;
}
}
return 0;
}
[-- Attachment #3: do_tcbm_att.txt --]
[-- Type: application/octet-stream, Size: 36 bytes --]
g++ -Wall -g -o tcbm_att tcbm_att.C
next reply other threads:[~2009-05-15 18:47 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-05-15 18:44 starlight [this message]
2009-05-18 16:36 ` Mel Gorman
-- strict thread matches above, loose matches on Subject: below --
2009-05-15 18:53 starlight
2009-05-20 11:35 ` Mel Gorman
2009-05-20 14:29 ` Mel Gorman
2009-05-20 14:53 ` Lee Schermerhorn
2009-05-20 15:05 ` Lee Schermerhorn
2009-05-20 15:41 ` Mel Gorman
2009-05-21 0:41 ` KOSAKI Motohiro
2009-05-22 16:41 ` Mel Gorman
2009-05-24 13:44 ` KOSAKI Motohiro
2009-05-25 8:51 ` Mel Gorman
2009-05-25 10:10 ` Hugh Dickins
2009-05-25 13:17 ` Mel Gorman
2009-05-15 5:32 starlight
2009-05-15 14:55 ` Mel Gorman
2009-05-15 15:02 ` starlight
[not found] <bug-13302-10286@http.bugzilla.kernel.org/>
2009-05-13 20:08 ` Andrew Morton
2009-05-14 10:53 ` Mel Gorman
2009-05-14 10:59 ` Mel Gorman
2009-05-14 17:20 ` starlight
2009-05-14 17:49 ` Mel Gorman
2009-05-14 18:42 ` starlight
2009-05-14 19:10 ` starlight
2009-05-14 17:16 ` starlight
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=6.2.5.6.2.20090515144058.03a55298@binnacle.cx \
--to=starlight@binnacle.cx \
--cc=agl@us.ibm.com \
--cc=akpm@linux-foundation.org \
--cc=bugme-daemon@bugzilla.kernel.org \
--cc=bugzilla-daemon@bugzilla.kernel.org \
--cc=ebmunson@us.ibm.com \
--cc=linux-mm@kvack.org \
--cc=mel@csn.ul.ie \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox