Statistics
| Branch: | Revision:

root / qemu-tech.texi @ e080e785

History | View | Annotate | Download (22.3 kB)

1
\input texinfo @c -*- texinfo -*-
2
@c %**start of header
3
@setfilename qemu-tech.info
4

    
5
@documentlanguage en
6
@documentencoding UTF-8
7

    
8
@settitle QEMU Internals
9
@exampleindent 0
10
@paragraphindent 0
11
@c %**end of header
12

    
13
@ifinfo
14
@direntry
15
* QEMU Internals: (qemu-tech).   The QEMU Emulator Internals.
16
@end direntry
17
@end ifinfo
18

    
19
@iftex
20
@titlepage
21
@sp 7
22
@center @titlefont{QEMU Internals}
23
@sp 3
24
@end titlepage
25
@end iftex
26

    
27
@ifnottex
28
@node Top
29
@top
30

    
31
@menu
32
* Introduction::
33
* QEMU Internals::
34
* Regression Tests::
35
* Index::
36
@end menu
37
@end ifnottex
38

    
39
@contents
40

    
41
@node Introduction
42
@chapter Introduction
43

    
44
@menu
45
* intro_features::        Features
46
* intro_x86_emulation::   x86 and x86-64 emulation
47
* intro_arm_emulation::   ARM emulation
48
* intro_mips_emulation::  MIPS emulation
49
* intro_ppc_emulation::   PowerPC emulation
50
* intro_sparc_emulation:: Sparc32 and Sparc64 emulation
51
* intro_other_emulation:: Other CPU emulation
52
@end menu
53

    
54
@node intro_features
55
@section Features
56

    
57
QEMU is a FAST! processor emulator using a portable dynamic
58
translator.
59

    
60
QEMU has two operating modes:
61

    
62
@itemize @minus
63

    
64
@item
65
Full system emulation. In this mode (full platform virtualization),
66
QEMU emulates a full system (usually a PC), including a processor and
67
various peripherals. It can be used to launch several different
68
Operating Systems at once without rebooting the host machine or to
69
debug system code.
70

    
71
@item
72
User mode emulation. In this mode (application level virtualization),
73
QEMU can launch processes compiled for one CPU on another CPU, however
74
the Operating Systems must match. This can be used for example to ease
75
cross-compilation and cross-debugging.
76
@end itemize
77

    
78
As QEMU requires no host kernel driver to run, it is very safe and
79
easy to use.
80

    
81
QEMU generic features:
82

    
83
@itemize
84

    
85
@item User space only or full system emulation.
86

    
87
@item Using dynamic translation to native code for reasonable speed.
88

    
89
@item
90
Working on x86, x86_64 and PowerPC32/64 hosts. Being tested on ARM,
91
HPPA, Sparc32 and Sparc64. Previous versions had some support for
92
Alpha and S390 hosts, but TCG (see below) doesn't support those yet.
93

    
94
@item Self-modifying code support.
95

    
96
@item Precise exceptions support.
97

    
98
@item The virtual CPU is a library (@code{libqemu}) which can be used
99
in other projects (look at @file{qemu/tests/qruncom.c} to have an
100
example of user mode @code{libqemu} usage).
101

    
102
@item
103
Floating point library supporting both full software emulation and
104
native host FPU instructions.
105

    
106
@end itemize
107

    
108
QEMU user mode emulation features:
109
@itemize
110
@item Generic Linux system call converter, including most ioctls.
111

    
112
@item clone() emulation using native CPU clone() to use Linux scheduler for threads.
113

    
114
@item Accurate signal handling by remapping host signals to target signals.
115
@end itemize
116

    
117
Linux user emulator (Linux host only) can be used to launch the Wine
118
Windows API emulator (@url{http://www.winehq.org}). A Darwin user
119
emulator (Darwin hosts only) exists and a BSD user emulator for BSD
120
hosts is under development. It would also be possible to develop a
121
similar user emulator for Solaris.
122

    
123
QEMU full system emulation features:
124
@itemize
125
@item
126
QEMU uses a full software MMU for maximum portability.
127

    
128
@item
129
QEMU can optionally use an in-kernel accelerator, like kvm. The accelerators 
130
execute some of the guest code natively, while
131
continuing to emulate the rest of the machine.
132

    
133
@item
134
Various hardware devices can be emulated and in some cases, host
135
devices (e.g. serial and parallel ports, USB, drives) can be used
136
transparently by the guest Operating System. Host device passthrough
137
can be used for talking to external physical peripherals (e.g. a
138
webcam, modem or tape drive).
139

    
140
@item
141
Symmetric multiprocessing (SMP) even on a host with a single CPU. On a
142
SMP host system, QEMU can use only one CPU fully due to difficulty in
143
implementing atomic memory accesses efficiently.
144

    
145
@end itemize
146

    
147
@node intro_x86_emulation
148
@section x86 and x86-64 emulation
149

    
150
QEMU x86 target features:
151

    
152
@itemize
153

    
154
@item The virtual x86 CPU supports 16 bit and 32 bit addressing with segmentation.
155
LDT/GDT and IDT are emulated. VM86 mode is also supported to run
156
DOSEMU. There is some support for MMX/3DNow!, SSE, SSE2, SSE3, SSSE3,
157
and SSE4 as well as x86-64 SVM.
158

    
159
@item Support of host page sizes bigger than 4KB in user mode emulation.
160

    
161
@item QEMU can emulate itself on x86.
162

    
163
@item An extensive Linux x86 CPU test program is included @file{tests/test-i386}.
164
It can be used to test other x86 virtual CPUs.
165

    
166
@end itemize
167

    
168
Current QEMU limitations:
169

    
170
@itemize
171

    
172
@item Limited x86-64 support.
173

    
174
@item IPC syscalls are missing.
175

    
176
@item The x86 segment limits and access rights are not tested at every
177
memory access (yet). Hopefully, very few OSes seem to rely on that for
178
normal use.
179

    
180
@end itemize
181

    
182
@node intro_arm_emulation
183
@section ARM emulation
184

    
185
@itemize
186

    
187
@item Full ARM 7 user emulation.
188

    
189
@item NWFPE FPU support included in user Linux emulation.
190

    
191
@item Can run most ARM Linux binaries.
192

    
193
@end itemize
194

    
195
@node intro_mips_emulation
196
@section MIPS emulation
197

    
198
@itemize
199

    
200
@item The system emulation allows full MIPS32/MIPS64 Release 2 emulation,
201
including privileged instructions, FPU and MMU, in both little and big
202
endian modes.
203

    
204
@item The Linux userland emulation can run many 32 bit MIPS Linux binaries.
205

    
206
@end itemize
207

    
208
Current QEMU limitations:
209

    
210
@itemize
211

    
212
@item Self-modifying code is not always handled correctly.
213

    
214
@item 64 bit userland emulation is not implemented.
215

    
216
@item The system emulation is not complete enough to run real firmware.
217

    
218
@item The watchpoint debug facility is not implemented.
219

    
220
@end itemize
221

    
222
@node intro_ppc_emulation
223
@section PowerPC emulation
224

    
225
@itemize
226

    
227
@item Full PowerPC 32 bit emulation, including privileged instructions,
228
FPU and MMU.
229

    
230
@item Can run most PowerPC Linux binaries.
231

    
232
@end itemize
233

    
234
@node intro_sparc_emulation
235
@section Sparc32 and Sparc64 emulation
236

    
237
@itemize
238

    
239
@item Full SPARC V8 emulation, including privileged
240
instructions, FPU and MMU. SPARC V9 emulation includes most privileged
241
and VIS instructions, FPU and I/D MMU. Alignment is fully enforced.
242

    
243
@item Can run most 32-bit SPARC Linux binaries, SPARC32PLUS Linux binaries and
244
some 64-bit SPARC Linux binaries.
245

    
246
@end itemize
247

    
248
Current QEMU limitations:
249

    
250
@itemize
251

    
252
@item IPC syscalls are missing.
253

    
254
@item Floating point exception support is buggy.
255

    
256
@item Atomic instructions are not correctly implemented.
257

    
258
@item There are still some problems with Sparc64 emulators.
259

    
260
@end itemize
261

    
262
@node intro_other_emulation
263
@section Other CPU emulation
264

    
265
In addition to the above, QEMU supports emulation of other CPUs with
266
varying levels of success. These are:
267

    
268
@itemize
269

    
270
@item
271
Alpha
272
@item
273
CRIS
274
@item
275
M68k
276
@item
277
SH4
278
@end itemize
279

    
280
@node QEMU Internals
281
@chapter QEMU Internals
282

    
283
@menu
284
* QEMU compared to other emulators::
285
* Portable dynamic translation::
286
* Condition code optimisations::
287
* CPU state optimisations::
288
* Translation cache::
289
* Direct block chaining::
290
* Self-modifying code and translated code invalidation::
291
* Exception support::
292
* MMU emulation::
293
* Device emulation::
294
* Hardware interrupts::
295
* User emulation specific details::
296
* Bibliography::
297
@end menu
298

    
299
@node QEMU compared to other emulators
300
@section QEMU compared to other emulators
301

    
302
Like bochs [3], QEMU emulates an x86 CPU. But QEMU is much faster than
303
bochs as it uses dynamic compilation. Bochs is closely tied to x86 PC
304
emulation while QEMU can emulate several processors.
305

    
306
Like Valgrind [2], QEMU does user space emulation and dynamic
307
translation. Valgrind is mainly a memory debugger while QEMU has no
308
support for it (QEMU could be used to detect out of bound memory
309
accesses as Valgrind, but it has no support to track uninitialised data
310
as Valgrind does). The Valgrind dynamic translator generates better code
311
than QEMU (in particular it does register allocation) but it is closely
312
tied to an x86 host and target and has no support for precise exceptions
313
and system emulation.
314

    
315
EM86 [4] is the closest project to user space QEMU (and QEMU still uses
316
some of its code, in particular the ELF file loader). EM86 was limited
317
to an alpha host and used a proprietary and slow interpreter (the
318
interpreter part of the FX!32 Digital Win32 code translator [5]).
319

    
320
TWIN [6] is a Windows API emulator like Wine. It is less accurate than
321
Wine but includes a protected mode x86 interpreter to launch x86 Windows
322
executables. Such an approach has greater potential because most of the
323
Windows API is executed natively but it is far more difficult to develop
324
because all the data structures and function parameters exchanged
325
between the API and the x86 code must be converted.
326

    
327
User mode Linux [7] was the only solution before QEMU to launch a
328
Linux kernel as a process while not needing any host kernel
329
patches. However, user mode Linux requires heavy kernel patches while
330
QEMU accepts unpatched Linux kernels. The price to pay is that QEMU is
331
slower.
332

    
333
The Plex86 [8] PC virtualizer is done in the same spirit as the now
334
obsolete qemu-fast system emulator. It requires a patched Linux kernel
335
to work (you cannot launch the same kernel on your PC), but the
336
patches are really small. As it is a PC virtualizer (no emulation is
337
done except for some privileged instructions), it has the potential of
338
being faster than QEMU. The downside is that a complicated (and
339
potentially unsafe) host kernel patch is needed.
340

    
341
The commercial PC Virtualizers (VMWare [9], VirtualPC [10], TwoOStwo
342
[11]) are faster than QEMU, but they all need specific, proprietary
343
and potentially unsafe host drivers. Moreover, they are unable to
344
provide cycle exact simulation as an emulator can.
345

    
346
VirtualBox [12], Xen [13] and KVM [14] are based on QEMU. QEMU-SystemC
347
[15] uses QEMU to simulate a system where some hardware devices are
348
developed in SystemC.
349

    
350
@node Portable dynamic translation
351
@section Portable dynamic translation
352

    
353
QEMU is a dynamic translator. When it first encounters a piece of code,
354
it converts it to the host instruction set. Usually dynamic translators
355
are very complicated and highly CPU dependent. QEMU uses some tricks
356
which make it relatively easily portable and simple while achieving good
357
performances.
358

    
359
After the release of version 0.9.1, QEMU switched to a new method of
360
generating code, Tiny Code Generator or TCG. TCG relaxes the
361
dependency on the exact version of the compiler used. The basic idea
362
is to split every target instruction into a couple of RISC-like TCG
363
ops (see @code{target-i386/translate.c}). Some optimizations can be
364
performed at this stage, including liveness analysis and trivial
365
constant expression evaluation. TCG ops are then implemented in the
366
host CPU back end, also known as TCG target (see
367
@code{tcg/i386/tcg-target.c}). For more information, please take a
368
look at @code{tcg/README}.
369

    
370
@node Condition code optimisations
371
@section Condition code optimisations
372

    
373
Lazy evaluation of CPU condition codes (@code{EFLAGS} register on x86)
374
is important for CPUs where every instruction sets the condition
375
codes. It tends to be less important on conventional RISC systems
376
where condition codes are only updated when explicitly requested. On
377
Sparc64, costly update of both 32 and 64 bit condition codes can be
378
avoided with lazy evaluation.
379

    
380
Instead of computing the condition codes after each x86 instruction,
381
QEMU just stores one operand (called @code{CC_SRC}), the result
382
(called @code{CC_DST}) and the type of operation (called
383
@code{CC_OP}). When the condition codes are needed, the condition
384
codes can be calculated using this information. In addition, an
385
optimized calculation can be performed for some instruction types like
386
conditional branches.
387

    
388
@code{CC_OP} is almost never explicitly set in the generated code
389
because it is known at translation time.
390

    
391
The lazy condition code evaluation is used on x86, m68k, cris and
392
Sparc. ARM uses a simplified variant for the N and Z flags.
393

    
394
@node CPU state optimisations
395
@section CPU state optimisations
396

    
397
The target CPUs have many internal states which change the way it
398
evaluates instructions. In order to achieve a good speed, the
399
translation phase considers that some state information of the virtual
400
CPU cannot change in it. The state is recorded in the Translation
401
Block (TB). If the state changes (e.g. privilege level), a new TB will
402
be generated and the previous TB won't be used anymore until the state
403
matches the state recorded in the previous TB. For example, if the SS,
404
DS and ES segments have a zero base, then the translator does not even
405
generate an addition for the segment base.
406

    
407
[The FPU stack pointer register is not handled that way yet].
408

    
409
@node Translation cache
410
@section Translation cache
411

    
412
A 16 MByte cache holds the most recently used translations. For
413
simplicity, it is completely flushed when it is full. A translation unit
414
contains just a single basic block (a block of x86 instructions
415
terminated by a jump or by a virtual CPU state change which the
416
translator cannot deduce statically).
417

    
418
@node Direct block chaining
419
@section Direct block chaining
420

    
421
After each translated basic block is executed, QEMU uses the simulated
422
Program Counter (PC) and other cpu state informations (such as the CS
423
segment base value) to find the next basic block.
424

    
425
In order to accelerate the most common cases where the new simulated PC
426
is known, QEMU can patch a basic block so that it jumps directly to the
427
next one.
428

    
429
The most portable code uses an indirect jump. An indirect jump makes
430
it easier to make the jump target modification atomic. On some host
431
architectures (such as x86 or PowerPC), the @code{JUMP} opcode is
432
directly patched so that the block chaining has no overhead.
433

    
434
@node Self-modifying code and translated code invalidation
435
@section Self-modifying code and translated code invalidation
436

    
437
Self-modifying code is a special challenge in x86 emulation because no
438
instruction cache invalidation is signaled by the application when code
439
is modified.
440

    
441
When translated code is generated for a basic block, the corresponding
442
host page is write protected if it is not already read-only. Then, if
443
a write access is done to the page, Linux raises a SEGV signal. QEMU
444
then invalidates all the translated code in the page and enables write
445
accesses to the page.
446

    
447
Correct translated code invalidation is done efficiently by maintaining
448
a linked list of every translated block contained in a given page. Other
449
linked lists are also maintained to undo direct block chaining.
450

    
451
On RISC targets, correctly written software uses memory barriers and
452
cache flushes, so some of the protection above would not be
453
necessary. However, QEMU still requires that the generated code always
454
matches the target instructions in memory in order to handle
455
exceptions correctly.
456

    
457
@node Exception support
458
@section Exception support
459

    
460
longjmp() is used when an exception such as division by zero is
461
encountered.
462

    
463
The host SIGSEGV and SIGBUS signal handlers are used to get invalid
464
memory accesses. The simulated program counter is found by
465
retranslating the corresponding basic block and by looking where the
466
host program counter was at the exception point.
467

    
468
The virtual CPU cannot retrieve the exact @code{EFLAGS} register because
469
in some cases it is not computed because of condition code
470
optimisations. It is not a big concern because the emulated code can
471
still be restarted in any cases.
472

    
473
@node MMU emulation
474
@section MMU emulation
475

    
476
For system emulation QEMU supports a soft MMU. In that mode, the MMU
477
virtual to physical address translation is done at every memory
478
access. QEMU uses an address translation cache to speed up the
479
translation.
480

    
481
In order to avoid flushing the translated code each time the MMU
482
mappings change, QEMU uses a physically indexed translation cache. It
483
means that each basic block is indexed with its physical address.
484

    
485
When MMU mappings change, only the chaining of the basic blocks is
486
reset (i.e. a basic block can no longer jump directly to another one).
487

    
488
@node Device emulation
489
@section Device emulation
490

    
491
Systems emulated by QEMU are organized by boards. At initialization
492
phase, each board instantiates a number of CPUs, devices, RAM and
493
ROM. Each device in turn can assign I/O ports or memory areas (for
494
MMIO) to its handlers. When the emulation starts, an access to the
495
ports or MMIO memory areas assigned to the device causes the
496
corresponding handler to be called.
497

    
498
RAM and ROM are handled more optimally, only the offset to the host
499
memory needs to be added to the guest address.
500

    
501
The video RAM of VGA and other display cards is special: it can be
502
read or written directly like RAM, but write accesses cause the memory
503
to be marked with VGA_DIRTY flag as well.
504

    
505
QEMU supports some device classes like serial and parallel ports, USB,
506
drives and network devices, by providing APIs for easier connection to
507
the generic, higher level implementations. The API hides the
508
implementation details from the devices, like native device use or
509
advanced block device formats like QCOW.
510

    
511
Usually the devices implement a reset method and register support for
512
saving and loading of the device state. The devices can also use
513
timers, especially together with the use of bottom halves (BHs).
514

    
515
@node Hardware interrupts
516
@section Hardware interrupts
517

    
518
In order to be faster, QEMU does not check at every basic block if an
519
hardware interrupt is pending. Instead, the user must asynchrously
520
call a specific function to tell that an interrupt is pending. This
521
function resets the chaining of the currently executing basic
522
block. It ensures that the execution will return soon in the main loop
523
of the CPU emulator. Then the main loop can test if the interrupt is
524
pending and handle it.
525

    
526
@node User emulation specific details
527
@section User emulation specific details
528

    
529
@subsection Linux system call translation
530

    
531
QEMU includes a generic system call translator for Linux. It means that
532
the parameters of the system calls can be converted to fix the
533
endianness and 32/64 bit issues. The IOCTLs are converted with a generic
534
type description system (see @file{ioctls.h} and @file{thunk.c}).
535

    
536
QEMU supports host CPUs which have pages bigger than 4KB. It records all
537
the mappings the process does and try to emulated the @code{mmap()}
538
system calls in cases where the host @code{mmap()} call would fail
539
because of bad page alignment.
540

    
541
@subsection Linux signals
542

    
543
Normal and real-time signals are queued along with their information
544
(@code{siginfo_t}) as it is done in the Linux kernel. Then an interrupt
545
request is done to the virtual CPU. When it is interrupted, one queued
546
signal is handled by generating a stack frame in the virtual CPU as the
547
Linux kernel does. The @code{sigreturn()} system call is emulated to return
548
from the virtual signal handler.
549

    
550
Some signals (such as SIGALRM) directly come from the host. Other
551
signals are synthetized from the virtual CPU exceptions such as SIGFPE
552
when a division by zero is done (see @code{main.c:cpu_loop()}).
553

    
554
The blocked signal mask is still handled by the host Linux kernel so
555
that most signal system calls can be redirected directly to the host
556
Linux kernel. Only the @code{sigaction()} and @code{sigreturn()} system
557
calls need to be fully emulated (see @file{signal.c}).
558

    
559
@subsection clone() system call and threads
560

    
561
The Linux clone() system call is usually used to create a thread. QEMU
562
uses the host clone() system call so that real host threads are created
563
for each emulated thread. One virtual CPU instance is created for each
564
thread.
565

    
566
The virtual x86 CPU atomic operations are emulated with a global lock so
567
that their semantic is preserved.
568

    
569
Note that currently there are still some locking issues in QEMU. In
570
particular, the translated cache flush is not protected yet against
571
reentrancy.
572

    
573
@subsection Self-virtualization
574

    
575
QEMU was conceived so that ultimately it can emulate itself. Although
576
it is not very useful, it is an important test to show the power of the
577
emulator.
578

    
579
Achieving self-virtualization is not easy because there may be address
580
space conflicts. QEMU user emulators solve this problem by being an
581
executable ELF shared object as the ld-linux.so ELF interpreter. That
582
way, it can be relocated at load time.
583

    
584
@node Bibliography
585
@section Bibliography
586

    
587
@table @asis
588

    
589
@item [1]
590
@url{http://citeseer.nj.nec.com/piumarta98optimizing.html}, Optimizing
591
direct threaded code by selective inlining (1998) by Ian Piumarta, Fabio
592
Riccardi.
593

    
594
@item [2]
595
@url{http://developer.kde.org/~sewardj/}, Valgrind, an open-source
596
memory debugger for x86-GNU/Linux, by Julian Seward.
597

    
598
@item [3]
599
@url{http://bochs.sourceforge.net/}, the Bochs IA-32 Emulator Project,
600
by Kevin Lawton et al.
601

    
602
@item [4]
603
@url{http://www.cs.rose-hulman.edu/~donaldlf/em86/index.html}, the EM86
604
x86 emulator on Alpha-Linux.
605

    
606
@item [5]
607
@url{http://www.usenix.org/publications/library/proceedings/usenix-nt97/@/full_papers/chernoff/chernoff.pdf},
608
DIGITAL FX!32: Running 32-Bit x86 Applications on Alpha NT, by Anton
609
Chernoff and Ray Hookway.
610

    
611
@item [6]
612
@url{http://www.willows.com/}, Windows API library emulation from
613
Willows Software.
614

    
615
@item [7]
616
@url{http://user-mode-linux.sourceforge.net/},
617
The User-mode Linux Kernel.
618

    
619
@item [8]
620
@url{http://www.plex86.org/},
621
The new Plex86 project.
622

    
623
@item [9]
624
@url{http://www.vmware.com/},
625
The VMWare PC virtualizer.
626

    
627
@item [10]
628
@url{http://www.microsoft.com/windowsxp/virtualpc/},
629
The VirtualPC PC virtualizer.
630

    
631
@item [11]
632
@url{http://www.twoostwo.org/},
633
The TwoOStwo PC virtualizer.
634

    
635
@item [12]
636
@url{http://virtualbox.org/},
637
The VirtualBox PC virtualizer.
638

    
639
@item [13]
640
@url{http://www.xen.org/},
641
The Xen hypervisor.
642

    
643
@item [14]
644
@url{http://kvm.qumranet.com/kvmwiki/Front_Page},
645
Kernel Based Virtual Machine (KVM).
646

    
647
@item [15]
648
@url{http://www.greensocs.com/projects/QEMUSystemC},
649
QEMU-SystemC, a hardware co-simulator.
650

    
651
@end table
652

    
653
@node Regression Tests
654
@chapter Regression Tests
655

    
656
In the directory @file{tests/}, various interesting testing programs
657
are available. They are used for regression testing.
658

    
659
@menu
660
* test-i386::
661
* linux-test::
662
* qruncom.c::
663
@end menu
664

    
665
@node test-i386
666
@section @file{test-i386}
667

    
668
This program executes most of the 16 bit and 32 bit x86 instructions and
669
generates a text output. It can be compared with the output obtained with
670
a real CPU or another emulator. The target @code{make test} runs this
671
program and a @code{diff} on the generated output.
672

    
673
The Linux system call @code{modify_ldt()} is used to create x86 selectors
674
to test some 16 bit addressing and 32 bit with segmentation cases.
675

    
676
The Linux system call @code{vm86()} is used to test vm86 emulation.
677

    
678
Various exceptions are raised to test most of the x86 user space
679
exception reporting.
680

    
681
@node linux-test
682
@section @file{linux-test}
683

    
684
This program tests various Linux system calls. It is used to verify
685
that the system call parameters are correctly converted between target
686
and host CPUs.
687

    
688
@node qruncom.c
689
@section @file{qruncom.c}
690

    
691
Example of usage of @code{libqemu} to emulate a user mode i386 CPU.
692

    
693
@node Index
694
@chapter Index
695
@printindex cp
696

    
697
@bye