1 |
Hi Beso, |
2 |
i've tried with your trip_points modification but it gives this error: |
3 |
|
4 |
bash: echo: write error: Invalid argument |
5 |
|
6 |
like i'm not able to write on that file. |
7 |
|
8 |
And of course acpitool gives me a random ACPI temperature: |
9 |
|
10 |
Battery #1 : charged |
11 |
AC adapter : on-line |
12 |
Thermal zone 1 : ok, 40 C |
13 |
|
14 |
PS: i followed all your suggestions concerning the microcode and fan |
15 |
option in the kernel. |
16 |
Could this be a problem of DSDT? |
17 |
|
18 |
regards, |
19 |
m |
20 |
|
21 |
On 10/10/07, Beso <givemesugarr@×××××.com> wrote: |
22 |
> |
23 |
> |
24 |
> 2007/10/10, Marco Calviani <marco.calviani@×××××.com>: |
25 |
> > Hi Beso, |
26 |
> > sorry i misunderstood your suggestion. I did what you suggest and |
27 |
> > this is the result of the trip_points: |
28 |
> > |
29 |
> > critical (S5): 100C |
30 |
> |
31 |
> wow.... you don't have anything that says to the cpu to slow down when it |
32 |
> reaches some point.... |
33 |
> now, to add some other trip points you have to copy these in a konsole with |
34 |
> root priviledges: |
35 |
> echo "passive: 78 C: tc1=3 tc2=1 tsp=150 devices=CPU0 \ |
36 |
> active[0]: 68 C: devices= FN1 \ |
37 |
> active[1]: 58 C: devices= FN2" >> |
38 |
> /proc/acpi/thermal_zone/TZ01/trip_points |
39 |
> |
40 |
> then do a cat on the thermal_zone/TZ0/trip_points to see if you've added the |
41 |
> lines for passive and active lines. |
42 |
> that means that when your thermal reaches 78 degrees it will slow down the |
43 |
> processor. from 58 to 68 it will turn on the fan but don't turn down the |
44 |
> speed of the processor, below 58 it will turn off the fan. |
45 |
> > |
46 |
> > |
47 |
> |
48 |
> > which i suppose is the reason why at that temperature the laptop switch |
49 |
> off. |
50 |
> > So, nothing except for the critical state. Should i have to add there |
51 |
> something? |
52 |
> |
53 |
> if the pc turns down then it can read from somewhere the actual thermal |
54 |
> point. you try to see after actually setting the things i've just said, if |
55 |
> your pc is behaving as it should. remember to also turn on the polling |
56 |
> frequency. without it it will not look for thermal changes. and remember to |
57 |
> actually compile the mce and speedstep features in the kernel and not as |
58 |
> module and reboot and then set the things i've mentioned. after that type |
59 |
> acpitool (it should be installed by default with the acpi package) and see |
60 |
> what it says. it should give something like this: |
61 |
> |
62 |
> > Battery #1 : charging, 46.00%, 01:17:04 |
63 |
> > AC adapter : on-line |
64 |
> > Thermal zone 1 : activ, 58 C |
65 |
> |
66 |
> it indicates, as you can see not only the battery and ac status but also the |
67 |
> current processor mode (active) and the current thermal temperature.... if |
68 |
> you don't have acpitool try acpi -t (you'll surely have either one or the |
69 |
> other) and it should indicate the thermal state and temperature. if this |
70 |
> command don't give you these infos then you'll have to be very careful using |
71 |
> your pc since acpi probably don't support santarosa well. you should then |
72 |
> unmask newer acpi in portage (adding acpi in /etc/package.keywords) and try |
73 |
> with the new acpi ( 1.0.6). |
74 |
> you may need some additional acpi modules, like ibm_acpi for example, but |
75 |
> that depends on your's pc brand. i'm looking around to see if there were |
76 |
> someone that had problems with santarosa and linux acpi, but for the moment |
77 |
> i couldn't find something useful. try what i've said and see if the things |
78 |
> work. if they work then append the tweak in some script that starts at boot |
79 |
> like the top of xdm script and you'll have a functional system. let me know |
80 |
> if you were succesful on that. |
81 |
> |
82 |
> |
83 |
> > An additional problem is this: |
84 |
> > doing a |
85 |
> > $ cat /proc/acpi/thermal_zone/TZ01/temperature |
86 |
> > it gives only a: |
87 |
> > |
88 |
> > temperature: 0C |
89 |
> > |
90 |
> > My question is: |
91 |
> > even if i change the polling frequency, how the fan can start if the |
92 |
> > temperature gives 0?? |
93 |
> > Do you know if it's possible to link the fan start with the core |
94 |
> > temperature instead of the ACPI thermal zone? |
95 |
> |
96 |
> probably the fan don't start since you have it as a module. i've curently |
97 |
> had the same problem which solved by compiling it integrated in the kernel. |
98 |
> the fan in your case should always be on, not always be off.... |
99 |
> |
100 |
> > |
101 |
> > Regards, |
102 |
> > m |
103 |
> > |
104 |
> > > |
105 |
> > > thats why i told you to do this commanda: |
106 |
> > > > echo "2 seconds" > |
107 |
> > > /proc/acpi/thermal_zone/TZ01/polling:frequency |
108 |
> > > > |
109 |
> > > this enables the polling of your thermal every 2 seconds. this should |
110 |
> be |
111 |
> > > enough. |
112 |
> > > do you have the other file that i mentioned: |
113 |
> > > > /proc/acpi/thermal_zone/TZ01/trip_points ?! |
114 |
> > > this sets the trip points for your processor. whitout it you governor |
115 |
> cannot |
116 |
> > > understand what to do even if it polls right. |
117 |
> > > |
118 |
> > > as for the kernel thigs, set these options: |
119 |
> > > select processor type: intel core2 instead of normal x86 |
120 |
> > > CONFIG_X86_SPEEDSTEP_CENTRINO=y instead of m |
121 |
> > > CONFIG_ACPI_FAN=y instead of m |
122 |
> > > CONFIG_MICROCODE=y instead of m (for what i know cpu micocode is needed |
123 |
> on |
124 |
> > > intels) |
125 |
> > > CONFIG_K8_NUMA=n instead of y (this should be the amdk8 numa, that you |
126 |
> > > should not need. if it's not then let him be) |
127 |
> > > CONFIG_X86_MCE_AMD=n instead of y (you don't need amd mce features since |
128 |
> > > they are not included into intel cpus) |
129 |
> > > CONFIG_NR_CPUS=32 <-- this leaves me a little dazzled: do you really |
130 |
> have 32 |
131 |
> > > cpus in your core?! for what i know this sets the real number of cpus |
132 |
> inside |
133 |
> > > the kernel, but i might be wrong. so if this is really what i think it |
134 |
> is, |
135 |
> > > ie the real nr of cpus (not virtual ones) set this to 2 or 4 based on |
136 |
> your |
137 |
> > > cpu cores. |
138 |
> > > CONFIG_HOTPLUG_CPU=n instead of y (i don't really think that you'll |
139 |
> unplug |
140 |
> > > your cpu from your laptop when the laptop is still running considering |
141 |
> that |
142 |
> > > you don't use multi cpus but a single multicore cpu. the same goes for |
143 |
> > > memory hotplug: i don't think that your laptop supports it, so just |
144 |
> disable |
145 |
> > > it.) |
146 |
> > > for what i have seen the acpi problems may be due to a failure in |
147 |
> loading |
148 |
> > > the intel speedstep module. if you look into the modules loaded (lsmod) |
149 |
> you |
150 |
> > > should not see it. so it's better to insert it directly in the kernel, |
151 |
> since |
152 |
> > > it is one of the first modules called (if you use it as a module you |
153 |
> should |
154 |
> > > be loading it with initramdisk before loading acpi to have a full acpi |
155 |
> > > configuration). |
156 |
> > > try setting these options and recompile and install the new kernel and |
157 |
> > > modules and reboot (kexec is not working on my amd turion with 64bit |
158 |
> enabled |
159 |
> > > and so may also be for your core2duo). |
160 |
> > > |
161 |
> > > 2007/10/10, Marco Calviani < marco.calviani@×××××.com>: |
162 |
> > > > Sorry i missed the attachment. |
163 |
> > > > |
164 |
> > > > regards, |
165 |
> > > > m |
166 |
> > > > |
167 |
> > > > |
168 |
> > > |
169 |
> > > |
170 |
> > > |
171 |
> > > -- |
172 |
> > > dott. ing. beso |
173 |
> > -- |
174 |
> > gentoo-laptop@g.o mailing list |
175 |
> > |
176 |
> > |
177 |
> |
178 |
> |
179 |
> |
180 |
> -- |
181 |
> dott. ing. beso |
182 |
-- |
183 |
gentoo-laptop@g.o mailing list |