Acer Aspire 1300 buggy BIOS - hardcoded patch

Acer Aspire 1300 buggy BIOS - hardcoded patch

[Luigi Belli]

[-- Attachment #1: Type: text/plain, Size: 443 bytes --]

I'm one of the owners of an Acer Aspire 1300 series with a buggy BIOS, so I 
decided to hardcode a patch with the correct PST table for my mobile Athlon 
XP 1800+.
I know very little of kernel internals, I only opened powernow-k7.c and tried 
to guess what had to be changed...
For me it works and I decided to post it here for those in my conditions 
waiting for a new bios or a newer cpufreq release.
It's a patch against kernel 2.4.21-ac4.

[-- Attachment #2: patch-aspire1300-1800+ --]
[-- Type: text/x-diff, Size: 1809 bytes --]

--- a/arch/i386/kernel/powernow-k7.c	2003-06-30 13:40:16.000000000 +0200
+++ b/arch/i386/kernel/powernow-k7.c	2003-06-30 13:41:19.000000000 +0200
@@ -167,6 +167,29 @@
 	unsigned int j, speed;
 	u8 fid, vid;

+/* hardcoded modifies for mobile Athlon xp 1800+ on Acer Aspire 1300 with buggy bios */
+
+	pst = kmalloc(10*sizeof(unsigned char), GFP_KERNEL);
+	if (!pst)
+		return -ENOMEM;
+
+	*pst++ = 4;  /*  666 MHz */
+	*pst++ = 19; /*   1.20 V */
+	*pst++ = 6;  /*  800 Mhz */
+	*pst++ = 19; /*   1.20 V */
+	*pst++ = 10; /* 1066 Mhz */
+	*pst++ = 19; /*   1.20 V */
+	*pst++ = 14; /* 1333 Mhz */
+	*pst++ = 11; /*   1.45 V */
+	*pst++ = 1;  /* 1533 Mhz */
+	*pst = 9;    /*   1.55 V */
+
+	pst -= 9;
+
+	fsb=133;
+
+/*************************************************/
+
 	powernow_table = kmalloc((sizeof(struct cpufreq_frequency_table) * (number_scales + 1)), GFP_KERNEL);
 	if (!powernow_table)
 		return -ENOMEM;
@@ -187,7 +210,7 @@
 #endif
 		}
 
-		dprintk (KERN_INFO PFX "   FID: 0x%x (%d.%dx [%dMHz])\t", fid,
+		dprintk (KERN_INFO PFX "   FID: 0x%x (%d.%dx [%dMHz])  ", fid,
 			fid_codes[fid] / 10, fid_codes[fid] % 10, speed);
 
 		if (speed < minimum_speed)
@@ -328,9 +351,11 @@
 				if ((etuple == pst->cpuid) && (maxfid==pst->maxfid) && (startvid==pst->startvid))
 				{
 					dprintk (KERN_INFO PFX "PST:%d (@%p)\n", i, pst);
-					dprintk (KERN_INFO PFX " cpuid: 0x%x\t", pst->cpuid);
-					dprintk ("fsb: %d\t", pst->fsbspeed);
-					dprintk ("maxFID: 0x%x\t", pst->maxfid);
+					dprintk (KERN_INFO PFX " cpuid: 0x%x  ", pst->cpuid);
+					/* modified for mobile Athlon XP 1800+ */
+					dprintk ("fsb: %d  ", 133);
+					/***************************************/
+					dprintk ("maxFID: 0x%x  ", pst->maxfid);
 					dprintk ("startvid: 0x%x\n", pst->startvid);
 
 					fsb = pst->fsbspeed;

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Ducrot Bruno]

On Mon, Jun 30, 2003 at 02:36:29PM +0200, Luigi Belli wrote:
> I'm one of the owners of an Acer Aspire 1300 series with a buggy BIOS, so I 
> decided to hardcode a patch with the correct PST table for my mobile Athlon 
> XP 1800+.
> I know very little of kernel internals, I only opened powernow-k7.c and tried 
> to guess what had to be changed...
> For me it works and I decided to post it here for those in my conditions 
> waiting for a new bios or a newer cpufreq release.
> It's a patch against kernel 2.4.21-ac4.


I am wondering if the _PCT table (from ACPI) are correct or not.
Could you send me a DSDT table, please?  For this, you have to
retrieve pmtools from intel site (this should be the latest link):

http://developer.intel.com/technology/iapc/acpi/downloads.htm


then go to pmtools-20010730/acpidmp/
make
./acpidmp DSDT > dsdt
gzip dsdt

then send me *privately* the dsdt.gz file.

I will then look if, at least, ACER guy give correct _PCT table.

Cheers,

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Stefan Gehn]

Hi,

On Monday June 30 2003 14:36, Luigi Belli wrote:
> I'm one of the owners of an Acer Aspire 1300 series with a buggy BIOS, so I
> decided to hardcode a patch with the correct PST table for my mobile Athlon
> XP 1800+.

Nice work :)

> I know very little of kernel internals, I only opened powernow-k7.c and
> tried to guess what had to be changed...
> For me it works and I decided to post it here for those in my conditions
> waiting for a new bios or a newer cpufreq release.
> It's a patch against kernel 2.4.21-ac4.

How to make this work for my xp 1500+?
I assume just stripping the 1533 Mhz setting and decreasing the length by 2 
should be enough, no? Otherwise, how to get the right voltages and 
multipliers for my CPU?

Bye, Stefan aka mETz
-- 
sgehn_AT_gmx.net | ICQ#51123152 | Moege der Pinguin mit euch sein

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Francesco Poli]

On Mon, 30 Jun 2003 14:36:29 +0200 Luigi Belli wrote:

> I'm one of the owners of an Acer Aspire 1300 series with a buggy BIOS,
> so I decided to hardcode a patch with the correct PST table for my
> mobile Athlon XP 1800+.

BTW, I have just installed a 2.4.21 + cpufreq patch 3 kernel on an Acer
Aspire 1302XV (AMD Mobile Athlon XP 1600+. It seems to work (the back
fan has stopped spinning so often and the case temperature looks lower).

Does anybody know a simple method to check if the BIOS is buggy or not?
Can I run some quick diagnostics?

-- 
    Francesco Poli       <frx AT firenze.linux.it>
======================================================
You're compiling a program and, all of a sudden, boom!
  -- from APT HOWTO, version 1.8.0

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Luigi Belli]

On Tuesday 01 July 2003 12:59 am, Francesco Poli wrote:

> Does anybody know a simple method to check if the BIOS is buggy or not?

Well... simply look at syslog and see if the frequencies are correct... for my 
BIOS my cpu has 5 steps: 4 at 500Mhz and 1 at 600Mhz, all at 1,4V.
That's obviously wrong...

Acer Aspire 1300 buggy BIOS - hardcoded patch

[Alex Young]

I have a similar problem with the bios of my sony vaio fx802. It has a mobile 
athlonxp 1500+ and has the following output from powernow-k7.

powernow: AMD K7 CPU detected.
powernow: PowerNOW! Technology present. Can scale: frequency and voltage.
powernow: Found PSB header at c00f74d0
powernow: Table version: 0x12
powernow: Flags: 0x0 (Mobile voltage regulator)
powernow: Settling Time: 100 microseconds.
powernow: Has 28 PST tables. (Only dumping ones relevant to this CPU).
powernow: PST:26 (@c00f76ac)
powernow:  cpuid: 0x780 fsb: 100        maxFID: 0x14    startvid: 0xe
powernow:    FID: 0x4 (5.0x [500MHz])   VID: 0xe (1.300V)
powernow:    FID: 0x8 (7.0x [700MHz])   VID: 0xe (1.300V)
powernow:    FID: 0xc (9.0x [900MHz])   VID: 0xe (1.300V)
powernow:    FID: 0x0 (11.0x [1100MHz]) VID: 0xe (1.300V)
powernow:    FID: 0x14 (13.0x [1300MHz])        VID: 0xe (1.300V)
powernow: Minimum speed 500 MHz. Maximum speed 1300 MHz.

Now my (possibly stupid) question is:
Why are the frequency/voltage settings controlled by the bios?
Surely these values are the same for all processors of the same speed and 
stepping number. Would it not be possible to obtain the recommended values 
directly from AMD and use them instead of the ones provided by the bios. Or 
can the motherboard hardware only generate certain values? 

Incidently does anybody have a athlonxp 1500+ with a 'correct' PST table? If 
so does voltage scaling make any noticeable difference to temperature or 
battery life?
-- 
Alex Young

RE: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Adachi, Kenichi]

Hi,
 
> Now my (possibly stupid) question is:
> Why are the frequency/voltage settings controlled by the bios?

It's probably because not all possible settings are necessarily
tolerable on all systems using same CPU.
System Vendors are quite careful as to which FID/VID settings they will
support on their product line.
Even if VRM is directly connected to CPU Core and thus configurable
independently of any other components on main board, still various
things such as TDP, power supply system, body size etc. must be taken
into acoount, I guess.

Thanks,
- Adachi, Kenichi

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Morrowind]

* martedì 01 luglio 2003, alle 00:59, Francesco Poli scrive:
> On Mon, 30 Jun 2003 14:36:29 +0200 Luigi Belli wrote:
> 
> BTW, I have just installed a 2.4.21 + cpufreq patch 3 kernel on an Acer
> Aspire 1302XV (AMD Mobile Athlon XP 1600+. It seems to work (the back
> fan has stopped spinning so often and the case temperature looks lower).
> 
> Does anybody know a simple method to check if the BIOS is buggy or not?
> Can I run some quick diagnostics?

Can you send the powernow-k7 debug info?
I've the same cpu and i need a correct pst table...

thanks


-- 
"A volte penso che la prova piu' sicura che esiste da qualche parte una
forma di vita intelligente e' il fatto che non ha mai tentato di mettersi
in contatto con noi."

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Ducrot Bruno]

On Mon, Jun 30, 2003 at 05:14:01PM +0200, Ducrot Bruno wrote:
> On Mon, Jun 30, 2003 at 02:36:29PM +0200, Luigi Belli wrote:
> > I'm one of the owners of an Acer Aspire 1300 series with a buggy BIOS, so I 
> > decided to hardcode a patch with the correct PST table for my mobile Athlon 
> > XP 1800+.
> > I know very little of kernel internals, I only opened powernow-k7.c and tried 
> > to guess what had to be changed...
> > For me it works and I decided to post it here for those in my conditions 
> > waiting for a new bios or a newer cpufreq release.
> > It's a patch against kernel 2.4.21-ac4.
> 
> 
> I am wondering if the _PCT table (from ACPI) are correct or not.
> Could you send me a DSDT table, please?  For this, you have to
> retrieve pmtools from intel site (this should be the latest link):
> 
> http://developer.intel.com/technology/iapc/acpi/downloads.htm
> 
> 
> then go to pmtools-20010730/acpidmp/
> make
> ./acpidmp DSDT > dsdt
> gzip dsdt
> 
> then send me *privately* the dsdt.gz file.
> 
> I will then look if, at least, ACER guy give correct _PCT table.
> 

Interresting.  From ACPI point-of-view, you have those states (I reproduce
only the different frequencies)

950 MHz
750 MHz
700 MHz
600 MHz
500 MHz

Ahem,  seems that not only your PST table is somehow wrong, but it seems
there is nothing to do with ACPI either, sorry.

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[davidendless]

> I'm one of the owners of an Acer Aspire 1300 series with a buggy
BIOS, so I
> decided to hardcode a patch with the correct PST table for my mobile
Athlon
> XP 1800+.

I tried the patch, and it seemed to work. Then I did some tests and I
found 
that in powersave (cat /proc/cpuinfo: cpu MHz 666.705) it runs faster
than at 
fullspeed (1533.422).

To test speed I used nbench and this simple script (in xterm at
fullscreen):

$ time for i in `seq 50000`; do echo "aaaa"; done

It took about 2,6 seconds without the powernow-k7 module loaded, the
same time 
with module loaded and powersave, and about 3,9 seconds with full
power and 
module loaded.

Isn't it strange?
Do you have the same behaviour?

David




--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Natsabe.it la più grande erboristeria multimarca online.
Solo prodotti di altissima qualità. 

Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=1308&d=1-7

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Luigi Belli]

On Tuesday 01 July 2003 02:24 pm, davidendless@email.it wrote:

> Isn't it strange?
> Do you have the same behaviour?

No, for me it works correctly, but after some testng I still experience lower 
temperatures under Windows XP.
However, it's better than nothing!

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Stefan Hoppe]

> No, for me it works correctly, but after some testing I still experience
> lower temperatures under Windows XP.
> However, it's better than nothing!

I'm still using rtdvs for frequency switching but the result should be the 
same. 

On Startup it tells me that it's on 1400MHz and 1450mV but when I switch to 
these values the temperature and powerconsumption increases. But when I 
switch to 1400MHz and 1350mV temperature and  power stays on the startupvalue 
(I posted it some month ago). So I use a Voltage 0.1V smaller than my Bios 
tells me for the max value and it's working. The temperature is the same as 
under Windows. I don't know if this is an rtdvs Problem, but I don't think 
so, because x86info gives me the Values I set. It's really strange.  So if 
you are using the hardcoded patch try to reduce the Voltage a litle bit to 
get the same values as under Windows.

Stefan

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Francesco Poli]

On Tue, 1 Jul 2003 01:20:18 +0200 Luigi Belli wrote:

> On Tuesday 01 July 2003 12:59 am, Francesco Poli wrote:
> 
> > Does anybody know a simple method to check if the BIOS is buggy or
> > not?
> 
> Well... simply look at syslog and see if the frequencies are
> correct... for my BIOS my cpu has 5 steps: 4 at 500Mhz and 1 at
> 600Mhz, all at 1,4V. That's obviously wrong...

Here's a /var/log/messages snippet:

AMD K7 CPU detected.
PowerNOW! Technology present. Can scale: frequency and voltage.
Found PSB header at c00e50e0
Table version: 0x12
Flags: 0x0 (Mobile voltage regulator)
Settling Time: 100 microseconds.
Has 1 PST tables. (Only dumping ones relevant to this CPU).
PST:0 (@c00e50f0)
 cpuid: 0x780^Ifsb: 100^ImaxFID: 0x16^Istartvid: 0xb
   FID: 0x4 (5.0x [500MHz])^IVID: 0x13 (1.200V)
   FID: 0x6 (6.0x [600MHz])^IVID: 0x13 (1.200V)
   FID: 0x8 (7.0x [700MHz])^IVID: 0x13 (1.200V)
   FID: 0xc (9.0x [900MHz])^IVID: 0x13 (1.200V)
   FID: 0xe (10.0x [1000MHz])^IVID: 0x11 (1.250V)
   FID: 0x0 (11.0x [1100MHz])^IVID: 0xe (1.300V)
   FID: 0x2 (12.0x [1200MHz])^IVID: 0xd (1.350V)
   FID: 0x16 (14.0x [1400MHz])^IVID: 0xb (1.450V)
Minimum speed 500 MHz. Maximum speed 1400 MHz.


The values look correct for an AMD Mobile Athlon XP 1600+, don't they?

BTW, a patch for the patch.
Well, actually, for its comments: you wrote Mhz somewhere and MHz
somewhere else. The right symbol is MHz, always!  ;-)


-- 
    Francesco Poli       <frx AT firenze.linux.it>
======================================================
You're compiling a program and, all of a sudden, boom!
  -- from APT HOWTO, version 1.8.0

Re: Acer Aspire 1300 buggy BIOS - hardcoded patch

[Ducrot Bruno]

> Interresting.  From ACPI point-of-view, you have those states (I reproduce
> only the different frequencies)
> 
> 950 MHz
> 750 MHz
> 700 MHz
> 600 MHz
> 500 MHz
> 
> Ahem,  seems that not only your PST table is somehow wrong, but it seems
> there is nothing to do with ACPI either, sorry.
> 

Ok.  Looking a little bit more, I think I can in fact retrieve
the exact same configuration (at least FID/VID), minus fsb and
settling time (which is 125us instead of 100us) than your override.

I am therefore wondering if that is OK to fallback to retrieve at
least those FID/VID via ACPI, but (ahem) still the fsb issue though.


The 5 states retrieved via _PSS look like this:

+---------+-------------------+---------------+------------+--------+
|  Freq.  | Power compsuption | settling time |   control  | verif. |
|---------+-------------------+---------------+------------+--------|
| 950 MHz |     24000 mW      |     125 us    | 0x009C418D | 0x018D |
| 750 MHz |     16337 mW      |     125 us    | 0x009C41C9 | 0x01C9 |
| 700 MHz |     15248 mW      |     125 us    | 0x009C41C8 | 0x01C8 |
| 600 MHz |     12084 mW      |     125 us    | 0x009C4226 | 0x0226 |
| 500 MHz |      9280 mW      |     125 us    | 0x009C4264 | 0x0264 |
+---------+-------------------+---------------+------------+--------+

Therefore, the 'control' stuff look like:
bits 0-4: FID
bits 5-9: VID
other bits: ?

rem: I must admit I don't know for what is the other bit-fields.
     Perhaps some may give real fsb and settling time?

which then permit to get those pairs (FID/VID):

state 0:  FID:0x0d (1263.5 MHz -- 950.0 MHz) / VID:0x0c (1.400)
state 1:  FID:0x09 (997.5 MHz -- 750.0 MHz) / VID:0x0e (1.300)
state 2:  FID:0x08 (931.0 MHz -- 700.0 MHz) / VID:0x0e (1.300)
state 3:  FID:0x06 (798.0 MHz -- 600.0 MHz) / VID:0x11 (1.250)
state 4:  FID:0x04 (665.0 MHz -- 500.0 MHz) / VID:0x13 (1.200)

rem: For the FID, I compute also the theorical freq. if fsb = 133 at
     first, then if fsb = 100.
and well, that look a little better than nothing.

Cheers,

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.