Re: hw.acpi.battery.life always at 100

On 18 avril at 12:36, Nate Lawson wrote:

> I'm adding -current to the CC since a change elsewhere in the kernel may 
> have triggered this.  Is anyone aware why it appears interrupts are now 
> disabled for a while just after launching init?  The change appeared in 
> 5-STABLE between March 17 and April 17.
> 
> Guy, can you try disabling usb?  Also, can you cvsup to dates between 
> March 17 and April 17 to try to narrow down the exact date this started 
> occurring?

  Hi all,


  Nate, I followed your instructions. First of all disabling USB
  didn't change anything. I then suped RELENG_5 to several dates in
  between March 17 and April 17 and found that the problem appeared
  after Mar-31-05:00 and before Mar-31-11:00. cvsup between these two
  timestamps shows two changes:

   Edit src/release/doc/en_US.ISO8859-1/relnotes/common/new.sgml
     Add delta 1.761.2.38 2005.03.31.06.28.58 hrs
   Edit src/sys/dev/acpica/acpivar.h
     Add delta 1.79.2.7 2005.03.31.06.03.59 njl

  and diffing acpivar.h reveals:

    _at__at_ -25,7 +25,7 _at__at_
      * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
      * SUCH DAMAGE.
      *
    - * $FreeBSD: src/sys/dev/acpica/acpivar.h,v 1.79.2.6 2005/03/02 09:18:41 obrien Exp $
    + * $FreeBSD: src/sys/dev/acpica/acpivar.h,v 1.79.2.7 2005/03/31 06:03:59 njl Exp $
      */
     
     #ifndef _ACPIVAR_H_
    _at__at_ -406,7 +406,7 _at__at_
     ACPI_HANDLE    acpi_GetReference(ACPI_HANDLE scope, ACPI_OBJECT *obj);
     
     #ifndef ACPI_MAX_THREADS
    -#define ACPI_MAX_THREADS       3
    +#define ACPI_MAX_THREADS       1
     #endif
     
     /* ACPI task kernel thread initialization. */

  I suped back to 5-STABLE and changed ACPI_MAX_THREADS accordingly
  and the problem has now disappeared:

  # sysctl hw.acpi.battery
  hw.acpi.battery.life: 87
  hw.acpi.battery.time: 167
  hw.acpi.battery.state: 1
  hw.acpi.battery.units: 1
  hw.acpi.battery.info_expire: 5

  # uname -a
  FreeBSD 5.4-STABLE FreeBSD 5.4-STABLE #7: Wed Apr 20 00:49:58 CEST
  2005     root_at_6nq.u-strasbg.fr:/usr/obj/usr/src/sys/GENERIC  i386

  BTW -CURRENT has the same problem (MFC from 1.91) and the same
  change fixes it.

  gb