File site/news/policy.de.md changed (mode: 100644) (index ca3be12..87d6ac6) |
... |
... |
exist, for example, the work done by Sam Zeloof and the Libre Silicon project: |
177 |
177 |
|
|
178 |
178 |
(Sam literally makes CPUs in his garage) |
(Sam literally makes CPUs in his garage) |
179 |
179 |
|
|
180 |
|
Why? |
|
181 |
|
==== |
|
|
180 |
|
More detailed insight about microcode |
|
181 |
|
===================================== |
|
182 |
|
|
|
183 |
|
To be clear: it is preferable that microcode be free. |
|
184 |
|
Not including CPU microcode updates is an absolute disaster for system |
|
185 |
|
stability and security, so Libreboot *includes microcode updates by default, in |
|
186 |
|
all modern release images, where possible to do so*. |
|
187 |
|
|
|
188 |
|
The CPU already has microcode burned into mask ROM. The microcode configures |
|
189 |
|
logic gates in the CPU, to implement an instruction set, via special *decoders* |
|
190 |
|
which are fixed-function; it is not possible, for example, to implement a RISCV |
|
191 |
|
ISA on an otherwise x86 processor. It is only possible for the microcode to |
|
192 |
|
implement x86, or *broken* x86, and the default microcode is almost always |
|
193 |
|
*broken x86* on Intel/AMD CPUs; it is inevitable, due to the complexity of |
|
194 |
|
these processors. |
|
195 |
|
|
|
196 |
|
These processors provide a way to supply microcode *updates*. These updates |
|
197 |
|
are volatile, and consequently must be applied during every boot cycle. The |
|
198 |
|
updates fix stability/reliability/security bugs, and their *absence* |
|
199 |
|
is *technically incorrect*, so you are strongly advised to install them. |
|
200 |
|
Examples of where these updates fix bugs: on ASUS KCMA-D8/KGPE-D16 |
|
201 |
|
and ThinkPad X200/T400/T500/W500/X200T/X200/R500/X301, the updates make |
|
202 |
|
hardware-based virtualization (via `kvm`) completely stable, where it would |
|
203 |
|
otherwise lead to a kernel panic. They allow those same thinkpads to be run with |
|
204 |
|
high CPU usage and I/O (RAM usage), without crashing (otherwise, it's very |
|
205 |
|
likely to encounter a kernel panic caused by a *Machine Check Exception*). |
|
206 |
|
|
|
207 |
|
Not including these updates will result in an unstable/undefined state. Intel |
|
208 |
|
themselves define which bugs affect which CPUs, and they define workarounds, or |
|
209 |
|
provide fixes in microcode. Based on this, software such as the Linux kernel |
|
210 |
|
can work around those bugs/quirks. Also, upstream versions of the Linux kernel |
|
211 |
|
can update the microcode at boot time (however, it is recommend still to do it |
|
212 |
|
from coreboot, for more stable memory controller initialization or “raminit”). |
|
213 |
|
Similar can be said about AMD CPUs. |
|
214 |
|
|
|
215 |
|
Once upon a time, Libreboot *excluded* microcode updates by default, but this |
|
216 |
|
lead to broken behaviour. Here are some examples: |
|
217 |
|
|
|
218 |
|
<https://browse.libreboot.org/lbmk.git/plain/resources/coreboot/default/patches/0012-fix-speedstep-on-x200-t400-Revert-cpu-intel-model_10.patch?id=9938fa14b1bf54db37c0c18bdfec051cae41448e> |
|
219 |
|
|
|
220 |
|
<https://browse.libreboot.org/lbmk.git/plain/resources/coreboot/default/patches/0018-Revert-cpu-intel-Configure-IA32_FEATURE_CONTROL-for-.patch?id=4b7be665968b67463ec36b9afc7e8736be0c9b51> |
|
221 |
|
|
|
222 |
|
These patches revert *bug fixes* in coreboot, fixes that happen to break other |
|
223 |
|
functionality but only when microcode updates are excluded. The most |
|
224 |
|
technically correct solution is to *not* apply the above patches, and instead |
|
225 |
|
supply microcode updates! |
|
226 |
|
|
|
227 |
|
You *need* microcode updates, or you will have a broken CPU; broken, because |
|
228 |
|
it literally behaves differently than it's supposed to, so software will have |
|
229 |
|
unpredictable bugs that could even cause data corruption - or worse. |
|
230 |
|
|
|
231 |
|
Why was this page written? |
|
232 |
|
========================== |
|
233 |
|
|
|
234 |
|
Many of the topics discussed here are actually hotly contested, by different |
|
235 |
|
sections of the free software movement. Libreboot has taken a firm stance. |
182 |
236 |
|
|
183 |
237 |
Firstly, observe the following graphic: |
Firstly, observe the following graphic: |
184 |
238 |
|
|
File site/news/policy.md changed (mode: 100644) (index dbb2661..59f9afb) |
... |
... |
exist, for example, the work done by Sam Zeloof and the Libre Silicon project: |
242 |
242 |
|
|
243 |
243 |
(Sam literally makes CPUs in his garage) |
(Sam literally makes CPUs in his garage) |
244 |
244 |
|
|
245 |
|
Why? |
|
246 |
|
==== |
|
|
245 |
|
More detailed insight about microcode |
|
246 |
|
===================================== |
|
247 |
|
|
|
248 |
|
To be clear: it is preferable that microcode be free. |
|
249 |
|
Not including CPU microcode updates is an absolute disaster for system |
|
250 |
|
stability and security, so Libreboot *includes microcode updates by default, in |
|
251 |
|
all modern release images, where possible to do so*. |
|
252 |
|
|
|
253 |
|
The CPU already has microcode burned into mask ROM. The microcode configures |
|
254 |
|
logic gates in the CPU, to implement an instruction set, via special *decoders* |
|
255 |
|
which are fixed-function; it is not possible, for example, to implement a RISCV |
|
256 |
|
ISA on an otherwise x86 processor. It is only possible for the microcode to |
|
257 |
|
implement x86, or *broken* x86, and the default microcode is almost always |
|
258 |
|
*broken x86* on Intel/AMD CPUs; it is inevitable, due to the complexity of |
|
259 |
|
these processors. |
|
260 |
|
|
|
261 |
|
These processors provide a way to supply microcode *updates*. These updates |
|
262 |
|
are volatile, and consequently must be applied during every boot cycle. The |
|
263 |
|
updates fix stability/reliability/security bugs, and their *absence* |
|
264 |
|
is *technically incorrect*, so you are strongly advised to install them. |
|
265 |
|
Examples of where these updates fix bugs: on ASUS KCMA-D8/KGPE-D16 |
|
266 |
|
and ThinkPad X200/T400/T500/W500/X200T/X200/R500/X301, the updates make |
|
267 |
|
hardware-based virtualization (via `kvm`) completely stable, where it would |
|
268 |
|
otherwise lead to a kernel panic. They allow those same thinkpads to be run with |
|
269 |
|
high CPU usage and I/O (RAM usage), without crashing (otherwise, it's very |
|
270 |
|
likely to encounter a kernel panic caused by a *Machine Check Exception*). |
|
271 |
|
|
|
272 |
|
Not including these updates will result in an unstable/undefined state. Intel |
|
273 |
|
themselves define which bugs affect which CPUs, and they define workarounds, or |
|
274 |
|
provide fixes in microcode. Based on this, software such as the Linux kernel |
|
275 |
|
can work around those bugs/quirks. Also, upstream versions of the Linux kernel |
|
276 |
|
can update the microcode at boot time (however, it is recommend still to do it |
|
277 |
|
from coreboot, for more stable memory controller initialization or “raminit”). |
|
278 |
|
Similar can be said about AMD CPUs. |
|
279 |
|
|
|
280 |
|
Once upon a time, Libreboot *excluded* microcode updates by default, but this |
|
281 |
|
lead to broken behaviour. Here are some examples: |
|
282 |
|
|
|
283 |
|
<https://browse.libreboot.org/lbmk.git/plain/resources/coreboot/default/patches/0012-fix-speedstep-on-x200-t400-Revert-cpu-intel-model_10.patch?id=9938fa14b1bf54db37c0c18bdfec051cae41448e> |
|
284 |
|
|
|
285 |
|
<https://browse.libreboot.org/lbmk.git/plain/resources/coreboot/default/patches/0018-Revert-cpu-intel-Configure-IA32_FEATURE_CONTROL-for-.patch?id=4b7be665968b67463ec36b9afc7e8736be0c9b51> |
|
286 |
|
|
|
287 |
|
These patches revert *bug fixes* in coreboot, fixes that happen to break other |
|
288 |
|
functionality but only when microcode updates are excluded. The most |
|
289 |
|
technically correct solution is to *not* apply the above patches, and instead |
|
290 |
|
supply microcode updates! |
|
291 |
|
|
|
292 |
|
You *need* microcode updates, or you will have a broken CPU; broken, because |
|
293 |
|
it literally behaves differently than it's supposed to, so software will have |
|
294 |
|
unpredictable bugs that could even cause data corruption - or worse. |
|
295 |
|
|
|
296 |
|
Why was this page written? |
|
297 |
|
========================== |
247 |
298 |
|
|
248 |
299 |
Firstly, observe the following graphic: |
Firstly, observe the following graphic: |
249 |
300 |
|
|
File site/news/policy.uk.md changed (mode: 100644) (index 5ab0b37..8832709) |
... |
... |
Libreboot вирішує цю ситуацію *суворо* та *принци |
171 |
171 |
|
|
172 |
172 |
(Сем буквально виробляє процесори в своєму гаражі) |
(Сем буквально виробляє процесори в своєму гаражі) |
173 |
173 |
|
|
174 |
|
Why? |
|
175 |
|
==== |
|
|
174 |
|
More detailed insight about microcode |
|
175 |
|
===================================== |
|
176 |
|
|
|
177 |
|
To be clear: it is preferable that microcode be free. |
|
178 |
|
Not including CPU microcode updates is an absolute disaster for system |
|
179 |
|
stability and security, so Libreboot *includes microcode updates by default, in |
|
180 |
|
all modern release images, where possible to do so*. |
|
181 |
|
|
|
182 |
|
The CPU already has microcode burned into mask ROM. The microcode configures |
|
183 |
|
logic gates in the CPU, to implement an instruction set, via special *decoders* |
|
184 |
|
which are fixed-function; it is not possible, for example, to implement a RISCV |
|
185 |
|
ISA on an otherwise x86 processor. It is only possible for the microcode to |
|
186 |
|
implement x86, or *broken* x86, and the default microcode is almost always |
|
187 |
|
*broken x86* on Intel/AMD CPUs; it is inevitable, due to the complexity of |
|
188 |
|
these processors. |
|
189 |
|
|
|
190 |
|
These processors provide a way to supply microcode *updates*. These updates |
|
191 |
|
are volatile, and consequently must be applied during every boot cycle. The |
|
192 |
|
updates fix stability/reliability/security bugs, and their *absence* |
|
193 |
|
is *technically incorrect*, so you are strongly advised to install them. |
|
194 |
|
Examples of where these updates fix bugs: on ASUS KCMA-D8/KGPE-D16 |
|
195 |
|
and ThinkPad X200/T400/T500/W500/X200T/X200/R500/X301, the updates make |
|
196 |
|
hardware-based virtualization (via `kvm`) completely stable, where it would |
|
197 |
|
otherwise lead to a kernel panic. They allow those same thinkpads to be run with |
|
198 |
|
high CPU usage and I/O (RAM usage), without crashing (otherwise, it's very |
|
199 |
|
likely to encounter a kernel panic caused by a *Machine Check Exception*). |
|
200 |
|
|
|
201 |
|
Not including these updates will result in an unstable/undefined state. Intel |
|
202 |
|
themselves define which bugs affect which CPUs, and they define workarounds, or |
|
203 |
|
provide fixes in microcode. Based on this, software such as the Linux kernel |
|
204 |
|
can work around those bugs/quirks. Also, upstream versions of the Linux kernel |
|
205 |
|
can update the microcode at boot time (however, it is recommend still to do it |
|
206 |
|
from coreboot, for more stable memory controller initialization or “raminit”). |
|
207 |
|
Similar can be said about AMD CPUs. |
|
208 |
|
|
|
209 |
|
Once upon a time, Libreboot *excluded* microcode updates by default, but this |
|
210 |
|
lead to broken behaviour. Here are some examples: |
|
211 |
|
|
|
212 |
|
<https://browse.libreboot.org/lbmk.git/plain/resources/coreboot/default/patches/0012-fix-speedstep-on-x200-t400-Revert-cpu-intel-model_10.patch?id=9938fa14b1bf54db37c0c18bdfec051cae41448e> |
|
213 |
|
|
|
214 |
|
<https://browse.libreboot.org/lbmk.git/plain/resources/coreboot/default/patches/0018-Revert-cpu-intel-Configure-IA32_FEATURE_CONTROL-for-.patch?id=4b7be665968b67463ec36b9afc7e8736be0c9b51> |
|
215 |
|
|
|
216 |
|
These patches revert *bug fixes* in coreboot, fixes that happen to break other |
|
217 |
|
functionality but only when microcode updates are excluded. The most |
|
218 |
|
technically correct solution is to *not* apply the above patches, and instead |
|
219 |
|
supply microcode updates! |
|
220 |
|
|
|
221 |
|
You *need* microcode updates, or you will have a broken CPU; broken, because |
|
222 |
|
it literally behaves differently than it's supposed to, so software will have |
|
223 |
|
unpredictable bugs that could even cause data corruption - or worse. |
|
224 |
|
|
|
225 |
|
Why was this page written? |
|
226 |
|
========================== |
176 |
227 |
|
|
177 |
228 |
Firstly, observe the following graphic: |
Firstly, observe the following graphic: |
178 |
229 |
|
|