summaryrefslogtreecommitdiff
path: root/2008/foss-silicon_manufacturer-elce2008/embedded-openexpo.mgp
blob: 59e30b7695de929f4c8a93557796bd33676d7364 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
%include "default.mgp"
%default 1 bgrad
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page
%nodefault
%back "blue"

%center
%size 7


How to do Embedded Linux [not] right

%center
%size 4
by

Harald Welte <laforge@gnumonks.org>

netfilter.org / openmoko.org / openpcd.org 
gpl-violations.org / openezx.org
hmw-consulting.de / viatech.com


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Introduction

Who is speaking to you?
		an independent Free Software developer, consultant and trainer
		13 years experience using/deploying and developing for Linux on server and workstation
		10 years professional experience doing Linux system + kernel level development 
		strong focus on network security and embedded
		expert in Free and Open Source Software (FOSS) copyright and licensing
		digital board-level hardware design, esp. embedded systems
		active developer and contributor to many FOSS projects
		thus, a techie, who will therefore not have fancy animated slides ;)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Introduction

Why am I qualified?
	The 'Linux community' POV
		Former kernel subsystem maintainer (netfilter/iptables)
		Initiator of OpenEZX project
		Author of various drivers for embedded hardware
	The 'embedded Linux done the rigth way' POV
		Lead System Architect (SW+HW) at Openmoko, Inc.
		Co-creator of Open Hardware + Software for RFID
			OpenPCD, OpenPICC, librfid, libmrtd
	The 'chip manufacturer' POV
		Open Source Liaison at VIA Technologies, Inc.
	The 'customer of consumer-grade embedded Linux' POV
		Done reverse-engineering on hundreds of devices for gpl-violations.org

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Linux is everywhere!

%image "linux_netfilter_singapore_entertainment.jpg"


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Linux is everywhere

Linux is everywhere!
	Linux mobile phones (Motorola, Openmoko)
	In-flight entertainment systems 
	Embedded networking gear
		DSLAMs
		rack monitoring
	Public payphones
	ATM's / PoS / vending machines
	Now even Fitness gear (Daum Ergometer)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Strengths of FOSS [0/4]


What are the true strengths of FOSS
	Innovative and creative development
	Security due to code review / bugreport / patches
	Long-term maintainable code
	Stable and reliable systems

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Strengths of FOSS [1/4]


Innovative and creative development
	easy-to-read existing codebase
	standard (FOSS) development tools
	thus, easy to modify and add features
	community will build around great new features/apps

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Strengths of FOSS [2/4]


Security due to code review / bugreport / patches
	all the code is out there
	many people are familiar with existing architecture
	code quality requirements usually very high
	community process allows quick and fast integration of bugfix

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Strengths of FOSS [3/4]


Long-term maintainable code, because
	there's a lot of attention on good software architecture
	many developers are familiar with the shared/common API's
	code quality requirements usually very high
	all code in mainline gets maintained and updated

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Strengths of FOSS [4/4]


Stable and reliable systems, because
	code quality requirements usually very high
	kernel releases are quite frequent
	all mainline code is automatically ported to new releases


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Reality Check


	So we should have the perfect world
		tons of embedded Linux products
		all of them maitainable, secure, stable
		encouraging lots of creative work on top of their codebase

	What is the reality
		tons of embedded Linux products
		none of the strengths of FOSS present in 99% of them

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Differences to PC Linux


Differences to Linux on a PC
	In the PC world, I can
		download the latest kernel from kernel.org
		compile + install it on almost every current+old platform
		have an almost 100% chance that it will boot and support all the major peripherals
		only some more obscure hardware might not have drivers yet
		update at any time to the latest 2.6.x.y maintenance release
		update at any time to the next 2.6.(x+1) release

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Differences to PC Linux


Differences to Linux on a PC
	In the PC world, I can
		take about any major Linux distribution, based on my own preference
		install and run that very distribution on about any hardware
		distribution kernels are very close to mainline these days
		benefit of regular security updates by distributions

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Differences to PC Linux


Differences to Linux on a PC
	In the Embedded world
		every CPU/SoC maker runs their own kernel tree
		often one kenrel tree per product, based on different mainline versions
		ages-old base revisions
		a neverending security nightmare
		no benefit from recent new features in mainline
		non-standard subsystems (e.g. different USB device or SDIO stack)
		proprietary drivers cause lock-in to old kernels

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Differences to PC Linux


Differences to Linux on a PC
	In the Embedded world
		I often do not have a choice of which distro to run
		There might actually be no distribution
		No regular security updates
		Often no package management for deploying those updates
		If there are distributions, they either need to use the kernel from the BSP (which is ages old) or creare yet another custom off-mainline branch/port

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Differences to PC Linux




%center
THIS SUCKS!


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
What does the vendor get


So what do the embedded vendors get?
	unstable software
	security nightmares
	unmaintainable code
	no innovation
	no user-contributed bug fixes

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
What does the vendor get


unstable software
	because the code really sucks in many cases
	because they patch around problems rather than solving them

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
What does the vendor get


security nightmares
	because they use stoneage forks of the kernel
	because they never contributed their code mainline
	because those forks can never be merged back with mainline again

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
What does the vendor get


unmaintainable code
	because they have one fork of the code per device (product)
	because their code quality sucks


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
What does the vendor get


no innovation
		because they try to hide their code (gpl-violations.org)
		because their R&D environment is non-standard
			weird cross-toolchains that nobody has seen before
			weird filesystems with custom patches that nobody knows
		because they add proprietary components to lock developers from adding features
			e.g. the entire web-based UI for embedded networking gear
			binary-only kernel modules that force people to use old kernels with no interesting new features
		because it is, overall, way too hard to develop on/for their platform
		because they don't disclose serial console and/or JTAG access

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Reality Check


So why do they still do it?
	there can only be one conclusion:
		they never understood the real potential of FOSS!
		all they do is try to compete with what proprietary competitors do
		they never think about creating platforms, every product is distinct/separate
		they have no interest in improving their products

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
What does the community get


So what does the community get?
	products that run some crappy fork of Linux somewhere under the hood
		but which we cannot really touch/modify without a lot of effort
	we face opposition from the product maker if we want to help him to improve

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
TODO (devica maker)


What should the device vendor do?
	stop thinking in terms of selling black boxes
	defining products that take advantage of the true strength of FOSS
	compete against the proprietary competition on a level that they can't match
	give up the idea of defining all aspects of an appliance
	rather think of building an extensible platforms and let community innovate

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
A note to chip vendors


There are two types of customers
	The Linux-aware customer
		understands FOSS much better than you do
		will share the criticism of this talk
		will likely go to a competitor who understands Linux better
	The Linux-unaware customer
		who just uses Linux to save per-unilt OS royalties
		who doesn't really care about most issues presented here
		who will create inferior products
	Linux-awareness is increasing, not decreasing
		now is already late, but if you don't have proper FOSS support on your agenda now, you will likely loose the "openness competition"

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
TODO (chip maker)


What should the chip vendors do?
	engange in "sustainable development"
		develop against latest mainline
		make your development trees public (use git!), don't just release stable snapshots as BSP to your tier-one customers
		actively interact with the community
		learn how to play by the rules (coding style, use common interfaces, no proprietary drivers)
		don't just do big code drops every now and then

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
TODO (chip maker)


What should the chip vendors do?
	Externally
		Don't mistake FOSS as just a technology.  It is a R&D philosophy!
		Provide public reference manuals with no NDA
		If you have to resort to NDA, make sure they are FOSS friendly
		Ensure you don't license IP cores that conflict with FOSS
	Internally
		Draft a proper in-house FOSS strategy with clear goals
		Don't expect your product mangers or engineers to know everything about FOSS without proper training
		Hire people with strong community background into your R&D and managemnt to facilitate the know-how transfer

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Lessons Learnt (chip maker)


	Chip Product Managers need to learn
		There's more Linux use with their chips than they ever learn
		Their customers are not just the tier-one customers
		If you want to support Linux, do it the mainline way.  If you support only N number of distributions, your 'N' will be growing and you'll be wasting R&D resources to support each one of them
		That there is no single 'contact window' / entity for Linux
		The big change is the FOSS development model, not the Linux API's
		Linux-aware customers care not only about performance+price, but also about the quality of the Linux port code


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Lessons Learnt (chip maker)


	BSP R&D Managers need to learn
		Linux is not just a set of API's for their developers to learn
		Linux is FOSS.  The FOSS R&D model is different
		Their engineers need to be encouraged to communicate
			and thus, need real internet (git, mailing lists, ..)
			have to get clear indication what is confidential and what not
		FOSS is about cooperation. You don't need to reinvent the wheel
			don't include your own 802.11 stack in your wifi driver
		If you do not merge your code mainline, you end up in a maintenance nightmare

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Lessons Learnt (chip maker)


	BSP Software Engineers need to learn
		Linux is not just a set of API's
		How and where to ask the right kind of questions
		How and where to look for the latest code
		Code is written to be read by other people, not just to execute


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
TODO (community)

What should the community do
	Provide non-partisan documentation
		on FOSS advantages, proper FOSS development
		hardware companies are intrested to learn, but don't know who to ask. If they ask a commercial distributor, then they get in bed with them, which is not the same as working with the mainline development community
	Something like a mentoring program
		take software/driver R&D by their hand and walk them through the mainline merge
	Don't scare them away
		They have to be taught about the communication / review culture
		Your valid criticism to patches has to be explained

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Problems (chipmaker)

	Patent licensing schemes
		e.g. MPEG patent license doesn't at all work in a FOSS model
	Technology licensing schemes
		e.g. Sony Memorystick. Impossible for a chip maker to provide FOSS driver
	Everyone in the industry has those very same problems
		Chip makers should cooperate to present their case together to the respective licensors

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Problems (chipmaker)

	Patent trolls
		openly accessible documentation invites patent trolls
	Why?
		patent trolls rarely read+understand FOSS code
		thus, open documentation increases the risk to be hit
	However
		this is a by-product of how the patent system currently works

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Outlook

Outlook
	We will see even more embedded Linux, e.g. Mobile phones
	We will see even more restricted devices (Tivo-ization, DRM for code)
		which go on _very_ thin ice even with GPLv2
		which almost completely remove all freedoms of FOSS
	We have some faint dim light at the end of a very far tunnel
		e.g. projects like Openmoko, who truly see openness as a feature
		e.g. chip makers who slowly open up a bit more
			on the PC side, Intel defiitely is setting the best overall example
			on the Embedded side, I see some movement in major players (TI, Marvell, Samsung, Infineon, ...)


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%page 
How to do Embedded Linux [not] right
Thanks

	Thanks for your attention.

	Some Time left for Q&A

	Enjoy the NLUUG + CELF ELCE 2008
personal git repositories of Harald Welte. Your mileage may vary