| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20091202142108.GA5815@elte.hu>
Date: Wed, 2 Dec 2009 15:21:08 +0100
From: Ingo Molnar <mingo@...e.hu>
To: Jan Beulich <JBeulich@...ell.com>
Cc: a.p.zijlstra@...llo.nl, tglx@...utronix.de,
torvalds@...ux-foundation.org, mingo@...hat.com, npiggin@...e.de,
linux-kernel@...r.kernel.org, linux-tip-commits@...r.kernel.org,
hpa@...or.com
Subject: Re: [tip:core/locking] locking, x86: Slightly shorten
__ticket_spin_trylock()
* Jan Beulich <JBeulich@...ell.com> wrote:
> >>> Ingo Molnar <mingo@...e.hu> 02.12.09 14:29 >>>
> >at first quick sight, this bit looks odd:
> >
> >+ union { int i; bool b; } new;
> >
> >+ return new.b;
> >
> >shouldnt that be short based, to work correctly in the 0-255 CPUs case?
>
> No, I can't see why. In both instances, we're using (and had been
> using previously, just with the added movzbl) the outcome of a setCC
> instruction, which produces valid bool (single byte) values. It is
> precisely that reason why I needed to introduce these unions, since
> the upper bytes of the register aren't valid (and shouldn't be looked
> at by the caller).
>
> Do you happen to have the vmlinux binary still around, to look at the
> code your compiler generated (I went through the code I got in quite a
> bit of detail to make sure it got translated correctly)?
I think the smoking gun is:
CONFIG_PARAVIRT_GUEST=y
CONFIG_PARAVIRT=y
CONFIG_PARAVIRT_SPINLOCKS=y
CONFIG_PARAVIRT_CLOCK=y
# CONFIG_PARAVIRT_DEBUG is not set
I disassembled the relevant bits for you:
792a9933 <_raw_spin_unlock>:
792a9933: 56 push %esi
792a9934: 53 push %ebx
792a9935: 89 c3 mov %eax,%ebx
792a9937: 81 78 04 ad 4e ad de cmpl $0xdead4ead,0x4(%eax)
792a993e: 74 0a je 792a994a <_raw_spin_unlock+0x17>
792a9940: ba 30 00 dd 79 mov $0x79dd0030,%edx
792a9945: e8 66 ff ff ff call 792a98b0 <spin_bug>
792a994a: 89 d8 mov %ebx,%eax
792a994c: ff 15 c0 6c f2 79 call *0x79f26cc0
792a9952: 85 c0 test %eax,%eax
792a9954: 75 0c jne 792a9962 <_raw_spin_unlock+0x2f>
792a9956: ba bf 00 dd 79 mov $0x79dd00bf,%edx
792a995b: 89 d8 mov %ebx,%eax
792a995d: e8 4e ff ff ff call 792a98b0 <spin_bug>
792a9962: 64 a1 84 a4 0a 7a mov %fs:0x7a0aa484,%eax
792a9968: 39 43 0c cmp %eax,0xc(%ebx)
792a996b: 74 0c je 792a9979 <_raw_spin_unlock+0x46>
792a996d: ba 3a 00 dd 79 mov $0x79dd003a,%edx
792a9972: 89 d8 mov %ebx,%eax
792a9974: e8 37 ff ff ff call 792a98b0 <spin_bug>
792a9979: 64 a1 e0 c0 0a 7a mov %fs:0x7a0ac0e0,%eax
792a997f: 39 43 08 cmp %eax,0x8(%ebx)
792a9982: 74 0c je 792a9990 <_raw_spin_unlock+0x5d>
792a9984: ba 46 00 dd 79 mov $0x79dd0046,%edx
792a9989: 89 d8 mov %ebx,%eax
792a998b: e8 20 ff ff ff call 792a98b0 <spin_bug>
792a9990: c7 43 0c ff ff ff ff movl $0xffffffff,0xc(%ebx)
792a9997: 89 d8 mov %ebx,%eax
792a9999: c7 43 08 ff ff ff ff movl $0xffffffff,0x8(%ebx)
792a99a0: ff 15 d4 6c f2 79 call *0x79f26cd4
792a99a6: 5b pop %ebx
792a99a7: 5e pop %esi
792a99a8: c3 ret
792a99a9 <_raw_spin_lock>:
792a99a9: 55 push %ebp
792a99aa: 57 push %edi
792a99ab: 56 push %esi
792a99ac: 53 push %ebx
792a99ad: 89 c3 mov %eax,%ebx
792a99af: 83 ec 10 sub $0x10,%esp
792a99b2: 81 78 04 ad 4e ad de cmpl $0xdead4ead,0x4(%eax)
792a99b9: 74 0a je 792a99c5 <_raw_spin_lock+0x1c>
792a99bb: ba 30 00 dd 79 mov $0x79dd0030,%edx
792a99c0: e8 eb fe ff ff call 792a98b0 <spin_bug>
792a99c5: 64 a1 84 a4 0a 7a mov %fs:0x7a0aa484,%eax
792a99cb: 39 43 0c cmp %eax,0xc(%ebx)
792a99ce: 75 0c jne 792a99dc <_raw_spin_lock+0x33>
792a99d0: ba a1 3e e0 79 mov $0x79e03ea1,%edx
792a99d5: 89 d8 mov %ebx,%eax
792a99d7: e8 d4 fe ff ff call 792a98b0 <spin_bug>
792a99dc: 64 a1 e0 c0 0a 7a mov %fs:0x7a0ac0e0,%eax
792a99e2: 39 43 08 cmp %eax,0x8(%ebx)
792a99e5: 75 0c jne 792a99f3 <_raw_spin_lock+0x4a>
792a99e7: ba 50 00 dd 79 mov $0x79dd0050,%edx
792a99ec: 89 d8 mov %ebx,%eax
792a99ee: e8 bd fe ff ff call 792a98b0 <spin_bug>
792a99f3: 89 d8 mov %ebx,%eax
792a99f5: ff 15 d0 6c f2 79 call *0x79f26cd0
792a99fb: 85 c0 test %eax,%eax
792a99fd: 0f 85 8c 00 00 00 jne 792a9a8f <_raw_spin_lock+0xe6>
792a9a03: 6b 15 78 55 f2 79 64 imul $0x64,0x79f25578,%edx
792a9a0a: bd 01 00 00 00 mov $0x1,%ebp
792a9a0f: 64 a1 84 a4 0a 7a mov %fs:0x7a0aa484,%eax
792a9a15: 89 44 24 04 mov %eax,0x4(%esp)
792a9a19: 05 0c 03 00 00 add $0x30c,%eax
792a9a1e: 89 df mov %ebx,%edi
792a9a20: 89 44 24 0c mov %eax,0xc(%esp)
792a9a24: 89 54 24 08 mov %edx,0x8(%esp)
792a9a28: 31 db xor %ebx,%ebx
792a9a2a: 31 f6 xor %esi,%esi
792a9a2c: eb 19 jmp 792a9a47 <_raw_spin_lock+0x9e>
792a9a2e: 89 f8 mov %edi,%eax
792a9a30: ff 15 d0 6c f2 79 call *0x79f26cd0
792a9a36: 85 c0 test %eax,%eax
792a9a38: 75 53 jne 792a9a8d <_raw_spin_lock+0xe4>
792a9a3a: b0 01 mov $0x1,%al
792a9a3c: e8 2f 5a ff ff call 7929f470 <__delay>
792a9a41: 83 c3 01 add $0x1,%ebx
792a9a44: 83 d6 00 adc $0x0,%esi
792a9a47: 83 fe 00 cmp $0x0,%esi
792a9a4a: 77 06 ja 792a9a52 <_raw_spin_lock+0xa9>
792a9a4c: 3b 5c 24 08 cmp 0x8(%esp),%ebx
792a9a50: 72 dc jb 792a9a2e <_raw_spin_lock+0x85>
792a9a52: 85 ed test %ebp,%ebp
792a9a54: 75 06 jne 792a9a5c <_raw_spin_lock+0xb3>
792a9a56: 31 db xor %ebx,%ebx
792a9a58: 31 f6 xor %esi,%esi
792a9a5a: eb f0 jmp 792a9a4c <_raw_spin_lock+0xa3>
792a9a5c: 57 push %edi
792a9a5d: 8b 54 24 08 mov 0x8(%esp),%edx
792a9a61: 31 ed xor %ebp,%ebp
792a9a63: 64 a1 e0 c0 0a 7a mov %fs:0x7a0ac0e0,%eax
792a9a69: ff b2 00 02 00 00 pushl 0x200(%edx)
792a9a6f: ff 74 24 14 pushl 0x14(%esp)
792a9a73: 50 push %eax
792a9a74: 68 d0 00 dd 79 push $0x79dd00d0
792a9a79: e8 8a 26 7d 00 call 79a7c108 <printk>
792a9a7e: e8 56 25 7d 00 call 79a7bfd9 <dump_stack>
792a9a83: e8 6e 81 d6 ff call 79011bf6 <arch_trigger_all_cpu_backtrace>
792a9a88: 83 c4 14 add $0x14,%esp
792a9a8b: eb 9b jmp 792a9a28 <_raw_spin_lock+0x7f>
792a9a8d: 89 fb mov %edi,%ebx
792a9a8f: 64 a1 e0 c0 0a 7a mov %fs:0x7a0ac0e0,%eax
792a9a95: 89 43 08 mov %eax,0x8(%ebx)
792a9a98: 64 a1 84 a4 0a 7a mov %fs:0x7a0aa484,%eax
792a9a9e: 89 43 0c mov %eax,0xc(%ebx)
792a9aa1: 83 c4 10 add $0x10,%esp
792a9aa4: 5b pop %ebx
792a9aa5: 5e pop %esi
792a9aa6: 5f pop %edi
792a9aa7: 5d pop %ebp
792a9aa8: c3 ret
792a9aa9: 90 nop
792a9aaa: 90 nop
792a9aab: 90 nop
79016a04 <__ticket_spin_lock>:
79016a04: ba 00 01 00 00 mov $0x100,%edx
79016a09: f0 66 0f c1 10 lock xadd %dx,(%eax)
79016a0e: 38 f2 cmp %dh,%dl
79016a10: 74 06 je 79016a18 <__ticket_spin_lock+0x14>
79016a12: f3 90 pause
79016a14: 8a 10 mov (%eax),%dl
79016a16: eb f6 jmp 79016a0e <__ticket_spin_lock+0xa>
79016a18: c3 ret
79016a19 <__ticket_spin_trylock>:
79016a19: 89 c2 mov %eax,%edx
79016a1b: 0f b7 02 movzwl (%edx),%eax
79016a1e: 38 e0 cmp %ah,%al
79016a20: 8d 88 00 01 00 00 lea 0x100(%eax),%ecx
79016a26: 75 05 jne 79016a2d <__ticket_spin_trylock+0x14>
79016a28: f0 66 0f b1 0a lock cmpxchg %cx,(%edx)
79016a2d: 0f 94 c1 sete %cl
79016a30: 88 c8 mov %cl,%al
79016a32: c3 ret
79016a33 <__ticket_spin_unlock>:
79016a33: fe 00 incb (%eax)
79016a35: c3 ret
79016a36 <__ticket_spin_is_locked>:
79016a36: 8b 10 mov (%eax),%edx
79016a38: 89 d0 mov %edx,%eax
79016a3a: c1 f8 08 sar $0x8,%eax
79016a3d: 31 d0 xor %edx,%eax
79016a3f: 84 c0 test %al,%al
79016a41: 0f 95 c0 setne %al
79016a44: 0f b6 c0 movzbl %al,%eax
79016a47: c3 ret
79016a48 <__ticket_spin_is_contended>:
79016a48: 8b 10 mov (%eax),%edx
79016a4a: 89 d0 mov %edx,%eax
79016a4c: c1 f8 08 sar $0x8,%eax
79016a4f: 29 d0 sub %edx,%eax
79016a51: 25 ff 00 00 00 and $0xff,%eax
79016a56: 48 dec %eax
79016a57: 0f 9f c0 setg %al
79016a5a: 0f b6 c0 movzbl %al,%eax
79016a5d: c3 ret
79016a5e <default_spin_lock_flags>:
79016a5e: ff 15 c8 6c f2 79 call *0x79f26cc8
79016a64: c3 ret
79016a65: 90 nop
79016a66: 90 nop
79016a67: 90 nop
I've also attached the config.
Ingo
View attachment "config" of type "text/plain" (76656 bytes)
Powered by blists - more mailing lists