libi386/biosacpi.c - bad RSDP checksum search

[Moore, Robert]

Part of the reason I asked the question is because the ACPICA Table Manager (the code that scans and extracts ACPI tables) has been specifically designed to be available at any time, even before virtual memory is available and before the rest of the ACPICA subsystem is initialized.

I haven't heard of this type of table scan before, but so be it. However, as you have seen, you are stuck with finding and fixing bugs that have been already resolved in the ACPICA code.

Bob


>-----Original Message-----
>From: Jung-uk Kim [mailto:jkim at FreeBSD.org]
>Sent: Wednesday, December 09, 2009 11:37 AM
>To: freebsd-acpi at FreeBSD.org
>Cc: Moore, Robert; Andrew Pantyukhin
>Subject: Re: libi386/biosacpi.c - bad RSDP checksum search
>
>On Wednesday 09 December 2009 01:46 pm, Moore, Robert wrote:
>> Might be a dumb question, but why is the bootloader looking around
>> for ACPI tables in the first place?
>
>FreeBSD/i386 bootloader loads ACPI kernel module if the root pointer
>is found and ACPI is not disabled by user.  Then, the pointer is
>passed to kernel as a hint for AcpiOsGetRootPointer() later.
>FreeBSD/amd64 just happened to share the same bootloader. :-)
>
>Jung-uk Kim
>
>> Thanks,
>> Bob
>>
>> >-----Original Message-----
>> >From: owner-freebsd-acpi at freebsd.org [mailto:owner-freebsd-
>> >acpi at freebsd.org] On Behalf Of John Baldwin
>> >Sent: Tuesday, December 08, 2009 4:50 AM
>> >To: freebsd-acpi at freebsd.org
>> >Cc: Andrew Pantyukhin
>> >Subject: Re: libi386/biosacpi.c - bad RSDP checksum search
>> >
>> >On Tuesday 08 December 2009 1:03:40 am Andrew Pantyukhin wrote:
>> >> Our boot loader stops searching memory at the first occurrence
>> >> of "RSD PTR" while there are BIOSes (e.g. some IBM System x)
>> >> that have multiple such strings and the first one does not
>> >> contain the correct checksum.
>> >>
>> >> The acpi-ca code does the right thing and continues the search.
>> >>
>> >> Any ACPI experts interested in fixing this? I'll be ready to
>> >> test.
>> >
>> >Are you sure?  It looks like it keeps going if the checksum fails.
>> >  Note the
>> >'continue' after the printf() about a bad checksum.
>> >
>> >/*
>> > * Find the RSDP in low memory.  See section 5.2.2 of the ACPI
>> > spec. */
>> >static ACPI_TABLE_RSDP *
>> >biosacpi_find_rsdp(void)
>> >{
>> >    ACPI_TABLE_RSDP	*rsdp;
>> >    uint16_t		*addr;
>> >
>> >    /* EBDA is the 1 KB addressed by the 16 bit pointer at 0x40E.
>> > */ addr = (uint16_t *)PTOV(0x40E);
>> >    if ((rsdp = biosacpi_search_rsdp((char *)(*addr << 4), 0x400))
>> > != NULL) return (rsdp);
>> >
>> >    /* Check the upper memory BIOS space, 0xe0000 - 0xfffff. */
>> >    if ((rsdp = biosacpi_search_rsdp((char *)0xe0000, 0x20000)) !=
>> > NULL) return (rsdp);
>> >
>> >    return (NULL);
>> >}
>> >
>> >static ACPI_TABLE_RSDP *
>> >biosacpi_search_rsdp(char *base, int length)
>> >{
>> >    ACPI_TABLE_RSDP	*rsdp;
>> >    u_int8_t		*cp, sum;
>> >    int			ofs, idx;
>> >
>> >    /* search on 16-byte boundaries */
>> >    for (ofs = 0; ofs < length; ofs += 16) {
>> >	rsdp = (ACPI_TABLE_RSDP *)PTOV(base + ofs);
>> >
>> >	/* compare signature, validate checksum */
>> >	if (!strncmp(rsdp->Signature, ACPI_SIG_RSDP,
>> > strlen(ACPI_SIG_RSDP))) {
>> >	    cp = (u_int8_t *)rsdp;
>> >	    sum = 0;
>> >	    for (idx = 0; idx < RSDP_CHECKSUM_LENGTH; idx++)
>> >		sum += *(cp + idx);
>> >	    if (sum != 0) {
>> >		printf("acpi: bad RSDP checksum (%d)\n", sum);
>> >		continue;
>> >	    }
>> >	    return(rsdp);
>> >	}
>> >    }
>> >    return(NULL);
>> >}
>> >
>> >--
>> >John Baldwin
>> >_______________________________________________
>> >freebsd-acpi at freebsd.org mailing list
>> >http://lists.freebsd.org/mailman/listinfo/freebsd-acpi
>> >To unsubscribe, send any mail to
>> > "freebsd-acpi-unsubscribe at freebsd.org"
>>
>> _______________________________________________
>> freebsd-acpi at freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-acpi
>> To unsubscribe, send any mail to
>> "freebsd-acpi-unsubscribe at freebsd.org"