Re: [PATCH][RFC] Proposal for a new watchdog interface using sysfs

From: Rusty Lynch (rusty@linux.co.intel.com)
Date: Fri Feb 14 2003 - 18:17:30 EST

Next message: Osamu Tomita: "Re: [PATCHSET] PC-9800 subarch. support for 2.5.60 (12/34) console"
Previous message: Ishikawa: "Subtle kernel configuration tool problem. Support for viper. Hi, I know that kernel configuration tool(s) are undergoing chnage, but here is one anomaly I noticed. I have an ABIT K7G raid motherboard, and attach an ATA/IDE disk to its non-raid controller. This board uses AMD chipset. I am using the kernel 2.4.20. (I use Debian GNU/Linux, but upgrade the kernel on my own regularly.) I notied earlier that the dmesg showed the following message: VP_IDE: IDE controller on PCI bus 00 dev 39 VP_IDE: detected chipset, but driver not compiled in! VP_IDE: chipset revision 6 VP_IDE: not 100% native mode: will probe irqs later After seeing this, I realized that I didn't enable the viper support and so I made sure that I have enabled the VIPER support in make xconfig Then I recompiled and installed the kernel for next boot, etc. However, no matter how I tried (and a few times), the message lines still appeared. (Actually, I have enabled the viper support a few weeks ago in make xconfig, and was suprised to find the message again, and this time I investigated a little on my own.) I re-checked the produced .config and was surprised to find there were NO mention of the supposedly enabled AMD viper support flag(!?). Just being curious, I DISABLED the viper support and saved the configuration under a different name and compared the file to the original config file. They were IDENTICAL! (This was done under make xconfig) (At this stage, I probably reenabled the viper support under make xconfig. Also, I ran make oldconfig just to be sure my .config is in sane state with respect to 2.4.20. But there were not discernable output, and make oldconfig finished without prompting input. At the end is the history of commands which I ran during this investigation.) Now I wasn't quite sure what to do, but then I tried the different configuration method, namely: make menuconfig What was really strange, is that the AMD VIPER support was marked [*] and so the tools seem to think that I have enabled it, but why not the proper line in .config file? (Are user preferences stored somewhere else other than the .config file itself ?) Now I was not trusting the tool(s) at this time very much, there must be some internal consistency problem. So I ticked off the [*] mark and saved it (under make menuconfig). I compared it with the old config produced by xconfig. Then I realized that diff output had lots of marked comment lines (probably, somehow the lines inside were re-ordered by make menuconfig or make oldconfig?). Anyway, AMD viper support was not in. Since I have tikced the mark off, this was to be expected. THEN, I ticked the AMD viper support in make menuconfig, and finally I found the AMD viper support flag line in .config: CONFIG_BLK_DEV_AMD74XX=y The above story is very strange, but it happened. For people working to improve the kernel configuration tool(s), I hope this post provides a data point for problem(s), which might face the people trying to configure the kernel. To people's credit, I think this is the first time something like this happened with the kernel configuration in my use of Linux for the last few years. Just to be complete, here is the command I ran while I tried to track down the problem and made sure the AMD74XX support is in. 106 make xconfig 107 dmesg grep VIPER 108 dmesg | grep VIPER 109 dmesg | grep Viper 110 dmesg | grep viper 111 dmesg | VP_IDE 112 dmesg | grep VP_IDE 113 grep AM /usr/src/linux/.config 114 cp ./config ./config.saved 115 cp ./.config ./config.saved 116 make xconfig 117 ls -ltr 118 diff -cibw ./.config new-config 119 make oldconfig 120 make xconfig 121 diff -cibw .config config.saved 122 make menuconfig 123 ls -ltr 124 diff -cibw .config new-config 125 diff -cibw .config new-config | grep -v ^# 126 diff -cibw .config new-config | grep -v "^! #" 127 diff -cibw .config new-config | grep -v "^[!+] #" 128 diff -cibw .config new-config | grep -v "^[!+] #" | more 129 make menuconfig 130 pwd 131 history 132 diff -cibw .config new-config | grep -v "^[!+] #" | more 133 make clean; make dep; make bzImage; make modules; make modules_install 134 history PS: it could be that my .config file contained some crufts left over from much earlier releases (2.3.x, 2.2.y, 2.0.z) and this may explain the problem."
In reply to: Pavel Machek: "Re: [PATCH][RFC] Proposal for a new watchdog interface using sysfs"
Next in thread: Alan Cox: "Re: [PATCH][RFC] Proposal for a new watchdog interface using sysfs"
Reply: Alan Cox: "Re: [PATCH][RFC] Proposal for a new watchdog interface using sysfs"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Fri, 2003-02-14 at 13:35, Pavel Machek wrote:
> Hi!
>
> > > > temperature (RO)
> > > > - show: prints temperature in degrees farenheit
> > >
> > > Please, use degrees celsius to keep it consistent with ACPI and
> > > lm-sensors.
> >
> > The ioctl interface is farenheit and has been since before Linux 2.0
> > That may not have been smart but we are stuck with it there at
> > least.
>
> Oops, that's bad.
>
> But I believe we should make it celsius in /sys, even if it means
> conversion somewhere.
>
> Pavel

The watchdog infrastructure would just show what ever integer the driver
provides via the watchdog_ops.get_temperature() function pointer, so it
would be up to the driver developer to decide if the data is really
Fahrenheit or whatever.

--rustyl

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Sat Feb 15 2003 - 22:00:59 EST