LinuxPPC-Dev Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
To: Eran Liberty <liberty@extricom.com>
Cc: linuxppc-dev@ozlabs.org, Steven Rostedt <srostedt@redhat.com>,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
	linux-kernel@vger.kernel.org,
	Steven Rostedt <rostedt@goodmis.org>
Subject: Re: ftrace introduces instability into kernel 2.6.27(-rc2,-rc3)
Date: Tue, 19 Aug 2008 10:42:11 -0400	[thread overview]
Message-ID: <20080819144211.GB21929@Krystal> (raw)
In-Reply-To: <48AAD702.10700@extricom.com>

* Eran Liberty (liberty@extricom.com) wrote:
> Mathieu Desnoyers wrote:
>> Can you also give us
>>
>> objdump -S --start-address=0xC00BB724 vmlinux | head 20
>>
>> ?
>>
>> Then we could compare the result with the OOPS instruction dump :
>>
>> 7c0802a6 bf61000c 3f60c038 7c3f0b78 90010024 7c7c1b78 7c9d2378 83db32a0
>> 73c00001 7f83e378 7fa4eb78 4082002f <00000000> 2f830000 409e0030 801b32a0
>>
>> Mathieu
>>
>>   
>
> to give you more context I have run:
> > powerpc-linux-gnu-objdump -S --start-address=0xC00BB700 vmlinux | head -n 
> 60
>
> the discrepancy starts at address:
> c00bb720: 40 82 00 30 <=> 40 82 00 2f
> c00bb724: 4b ff fe 61 <=> 00 00 00 00
>

Ah !

I think I see what could be wrong :

First we have :

static unsigned int ftrace_nop = 0x60000000;

We probably replace the original function call by this nop.

Then we do :

notrace unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
{
        static unsigned int op;

        /*
         * It would be nice to just use create_function_call, but that will
         * update the code itself. Here we need to just return the
         * instruction that is going to be modified, without modifying the
         * code.
         */
        addr = GET_ADDR(addr);

        /* Set to "bl addr" */
        op = 0x48000001 | (ftrace_calc_offset(ip, addr) & 0x03fffffc);

        /*
         * No locking needed, this must be called via kstop_machine
         * which in essence is like running on a uniprocessor machine.
         */
        return (unsigned char *)&op;
}

And I guess we must be doing a 0x48000001 | 0x0; or something ?

Also, we have to consider that POWERPC 64 functions are :

/* PowerPC64's functions are data that points to the functions */

And this does not seem to hold for ppc32. Therefore, it is strange to me
that the same code is used for the update.. are we updating the correct
site ?

Mathieu

> vmlinux:     file format elf32-powerpc
>
> Disassembly of section .text:
>
> c00bb700 <d_lookup+0x1c>:
> * d_lookup() is protected against the concurrent renames in some unrelated
> * directory using the seqlockt_t rename_lock.
> */
>
> struct dentry * d_lookup(struct dentry * parent, struct qstr * name)
> {
> c00bb700:       7c 3f 0b 78     mr      r31,r1
> c00bb704:       90 01 00 24     stw     r0,36(r1)
> c00bb708:       7c 7c 1b 78     mr      r28,r3
> c00bb70c:       7c 9d 23 78     mr      r29,r4
> c00bb710:       83 db 32 a0     lwz     r30,12960(r27)
> {
>        unsigned ret;
>
> repeat:
>        ret = sl->sequence;
>        smp_rmb();
> c00bb714:       73 c0 00 01     andi.   r0,r30,1
>        struct dentry * dentry = NULL;
>        unsigned long seq;
>
>        do {
>                seq = read_seqbegin(&rename_lock);
>                dentry = __d_lookup(parent, name);
> c00bb718:       7f 83 e3 78     mr      r3,r28
> c00bb71c:       7f a4 eb 78     mr      r4,r29
>        if (unlikely(ret & 1)) {
> c00bb720:       40 82 00 30     bne-    c00bb750 <d_lookup+0x6c>
> c00bb724:       4b ff fe 61     bl      c00bb584 <__d_lookup>
>                if (dentry)
> c00bb728:       2f 83 00 00     cmpwi   cr7,r3,0
> c00bb72c:       40 9e 00 30     bne-    cr7,c00bb75c <d_lookup+0x78>
> *
> * If sequence value changed then writer changed data while in section.
> */
> static __always_inline int read_seqretry(const seqlock_t *sl, unsigned 
> start)
> {
>        smp_rmb();
> c00bb730:       80 1b 32 a0     lwz     r0,12960(r27)
>                        break;
>        } while (read_seqretry(&rename_lock, seq));
> c00bb734:       7f 80 f0 00     cmpw    cr7,r0,r30
> c00bb738:       41 9e 00 24     beq-    cr7,c00bb75c <d_lookup+0x78>
> /* Start of read calculation -- fetch last complete writer token */
> static __always_inline unsigned read_seqbegin(const seqlock_t *sl)
> {
>        unsigned ret;
>
> repeat:
> c00bb73c:       7c 1e 03 78     mr      r30,r0
>        ret = sl->sequence;
>        smp_rmb();
> c00bb740:       73 c0 00 01     andi.   r0,r30,1
>        struct dentry * dentry = NULL;
>

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

  reply	other threads:[~2008-08-19 14:42 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <48591941.4070408@extricom.com>
     [not found] ` <48A92E15.2080709@extricom.com>
2008-08-18 15:07   ` ftrace introduces instability into kernel 2.6.27(-rc2,-rc3) Steven Rostedt
2008-08-18 15:47     ` Mathieu Desnoyers
2008-08-18 16:12       ` Steven Rostedt
2008-08-18 17:04         ` Mathieu Desnoyers
2008-08-18 17:21       ` Scott Wood
2008-08-18 18:27         ` Steven Rostedt
2008-08-18 18:29           ` Scott Wood
2008-08-19  1:53           ` Benjamin Herrenschmidt
2008-08-19  2:28             ` Steven Rostedt
2008-08-19  2:39               ` Benjamin Herrenschmidt
2008-08-19  2:41                 ` Steven Rostedt
2008-08-19  2:47                   ` Mathieu Desnoyers
2008-08-19  3:32                     ` Steven Rostedt
2008-08-19  3:36                       ` Mathieu Desnoyers
2008-08-19  4:00                         ` Steven Rostedt
2008-08-19 16:47                     ` Steven Rostedt
2008-08-19 17:34                       ` Mathieu Desnoyers
2008-08-19 21:08                         ` Steven Rostedt
2008-08-20  9:40                           ` Nick Piggin
2008-08-19 21:47                         ` Benjamin Herrenschmidt
2008-08-19 23:58                           ` Jeremy Fitzhardinge
2008-08-20  1:17                             ` Benjamin Herrenschmidt
2008-08-19  2:56                 ` Benjamin Herrenschmidt
2008-08-19  3:12                   ` Steven Rostedt
2008-08-19  4:17                     ` Benjamin Herrenschmidt
2008-08-20  7:18                       ` Benjamin Herrenschmidt
2008-08-20 13:14                         ` Steven Rostedt
2008-08-20 13:19                           ` Steven Rostedt
2008-08-20 13:36                             ` Eran Liberty
2008-08-20 13:43                               ` Steven Rostedt
2008-08-20 14:02                                 ` Eran Liberty
2008-08-20 14:55                                   ` Jon Smirl
2008-08-20 15:23                                     ` Steven Rostedt
2008-08-20 18:23                                     ` Eran Liberty
2008-08-20 18:33                                       ` Steven Rostedt
2008-08-20 15:27                                   ` Steven Rostedt
2008-08-20 21:37                                   ` Benjamin Herrenschmidt
2008-08-20 14:16                           ` Josh Boyer
2008-08-20 14:22                             ` Steven Rostedt
2008-08-20 14:50                               ` Josh Boyer
2008-08-20 21:36                           ` Benjamin Herrenschmidt
2008-08-20 21:44                             ` Steven Rostedt
2008-08-18 18:47         ` Steven Rostedt
2008-08-18 18:56           ` Scott Wood
2008-08-18 19:28             ` Steven Rostedt
2008-08-18 18:25     ` Eran Liberty
2008-08-18 18:41       ` Mathieu Desnoyers
2008-08-19  1:54         ` Benjamin Herrenschmidt
2008-08-19  9:56         ` Eran Liberty
2008-08-19 13:02           ` Mathieu Desnoyers
2008-08-19 21:46             ` Benjamin Herrenschmidt
2008-08-18 18:50       ` Steven Rostedt
2008-08-19 12:09         ` Eran Liberty
2008-08-19 13:05           ` Mathieu Desnoyers
2008-08-19 14:21             ` Eran Liberty
2008-08-19 14:42               ` Mathieu Desnoyers [this message]
2008-08-19 20:15           ` Steven Rostedt
2008-08-20 11:18             ` Eran Liberty
2008-08-20 13:12               ` Steven Rostedt
2008-08-19  1:51     ` Benjamin Herrenschmidt

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=20080819144211.GB21929@Krystal \
    --to=mathieu.desnoyers@polymtl.ca \
    --cc=liberty@extricom.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linuxppc-dev@ozlabs.org \
    --cc=paulmck@linux.vnet.ibm.com \
    --cc=rostedt@goodmis.org \
    --cc=srostedt@redhat.com \
    /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