Skip to contentSkip to conversation

kernel update (relative to 2.4.22)

DISCUSSION Bjorn Helgaas 2 participants 3 messages Sep 9, 2003 – Sep 12, 2003 ACPI

lore
Share & tools
lore
#0 lore raw

The ia64 kernel patch for Linux 2.4.22 is available here:

    ftp://ftp.kernel.org/pub/linux/kernel/ports/ia64/v2.4/linux-2.4.22-ia64-030909.diff.gz

The current 2.4-based ia64 tree is also available as a BitKeeper repository. You can browse the changelog and the source files with a normal browser at:

    http://lia64.bkbits.net:8080/linux-ia64-2.4

or you can use the BitKeeper tools to maintain a local copy of the tree like this:

    $ bk clone bk://lia64.bkbits.net/linux-ia64-2.4 linux-ia64-2.4

Bjorn

Changes since the 2.4.21-ia64-030702 patch:

* IA64:
- kernel entry/exit optimizations (Rohit Seth).
- add support for 16-byte instructions (CSD/SSD) (Rohit Seth).
- get supported TLB purge page sizes from PAL (Rohit Seth).
- use "hint @pause" in cpu_relax() and spinlock contention path (Rohit Seth).
- add barrier to cpu_relax() (Bjorn Helgaas).
- use PAL_HALT_LIGHT in cpu_idle() (disabled in generic config to avoid firmware defects) (Alex Williamson).
- improve scaling of fpassist printk (John Hawkes).
- fix EFI memory map trimming (Tony Luck).
* I/O:
- return PCI domain for pci_controller_num() (Matthew Wilcox).
- add/cleanup HP PCI device names (Alex Williamson).
- fix AGP ACPI device claim (Bjorn Helgaas).
- add HP AGP 8X support (Bjorn Helgaas).
* ACPI:
- new ACPI CA version 20030813! (from upstream).
- fix ACPI interrupt polarity/trigger interpretation (Takayoshi Kochi).
- fix ACPI interrupt to IRQ mapping (H. J. Lu).
- remove unused acpi_pci_get_translations() (Bjorn Helgaas).
- add ACPI serial port discovery error checking (Bjorn Helgaas).
- add ACPI serial port discovery for devices in IO port space (Bjorn Helgaas).
- prefer X_DSDT over DSDT, if it exists (Alex Williamson).
- handle more than NR_CPUS gracefully from 2.5 (Matt Chapman).
* MCA/INIT/CMC/CPE:
- rework CMC/CPE polling using cascaded interrupts (Alex Williamson).
- backtrace all processes on INIT from 2.5 (David Mosberger).
- clean up INIT code (Tony Luck).
- make OS_INIT work again (from 2.5) (Tony Luck).
- pass GP physical address to SAL (Keith Owens).
- find correct offset of OEM MCA data (Keith Owens).
* Perfmon:
- fix rum/rsm usage (Stephane Eranian).
* IA32 emulation:
- fix F_GETLK64/F_SETLK64 handling (Arun Sharma).
- fix FP denormal handling (Arun Sharma).
- fix memory leak in sys32_execve (Tony Luck).
* SGI SN and NUMA support:
- add NUMA support (Jesse Barnes, Takayoshi Kochi, Jack Steiner, others).
- support for non-identity-mapped kernels (Jack Steiner).
- misc SGI fixes (Jack Steiner).
- another SGI update (Erik Jacobson).
* Misc bug-fixes:
- cleanup /dev/mem cacheability checking from 2.5 (David Mosberger).
- unwind UNW_DPRINT() typo fixes (Keith Owens).
- user stack size configurable via RLIMIT_STACK from 2.5 (Martin Hicks).
* Misc cleanup:
- fix CONFIG_SERIAL_HCDP documentation and indentation (Bjorn Helgaas).
- fix hugetlb compile problem with gcc-3.4 (Stephane Eranian).
- configs updated, added new "numa" generic config (Bjorn Helgaas).

#1 lore raw

Hi Bjorn,

For NUMA configuration, we need the attached patch which is the same fix as 2.6. I confirmed both the DIG and GENERIC kernels boot on 8-way (2-node) NEC TX7.

As for drivers/acpi/tables.c part, I think Andrew or Len is already aware.

From: Bjorn Helgaas <[email protected]>
Subject: kernel update (relative to 2.4.22)
Date: Tue, 9 Sep 2003 11:28:32 -0600

Quoted 5 lines

* SGI SN and NUMA support:
- add NUMA support (Jesse Barnes, Takayoshi Kochi, Jack Steiner, others).
- support for non-identity-mapped kernels (Jack Steiner).
- misc SGI fixes (Jack Steiner).
- another SGI update (Erik Jacobson).

Takayoshi Kochi

#2 lore raw
In reply to Bjorn Helgaas

From: Takayoshi Kochi <[email protected]>
Subject: Re: kernel update (relative to 2.4.22)
Date: Fri, 12 Sep 2003 21:43:19 +0900 (JST)

Quoted 5 lines

Hi Bjorn,

For NUMA configuration, we need the attached patch which is the same fix as 2.6. I confirmed both the DIG and GENERIC kernels boot on 8-way (2-node) NEC TX7.

Sorry, I forgot the attachment.

Takayoshi Kochi

drivers/acpi/tables.c renamed from drivers/acpi/tables.c.save +1 -13
@@ -69,7 +69,7 @@
 
 static unsigned long		sdt_pa;		/* Physical Address */
 static unsigned long		sdt_count;	/* Table count */
-static struct acpi_table_sdt	*sdt_entry;
+static struct acpi_table_sdt	sdt_entry[ACPI_MAX_TABLES];
 
 void
 acpi_table_print (
@@ -425,12 +425,6 @@
 			sdt_count = ACPI_MAX_TABLES;
 		}
 
-		sdt_entry = alloc_bootmem(sdt_count * sizeof(struct acpi_table_sdt));
-		if (!sdt_entry) {
-			printk(KERN_ERR "ACPI: Could not allocate mem for SDT entries!\n");
-			return -ENOMEM;
-		}
-
 		for (i = 0; i < sdt_count; i++)
 			sdt_entry[i].pa = (unsigned long) mapped_xsdt->entry[i];
 	}
@@ -477,12 +471,6 @@
 			sdt_count = ACPI_MAX_TABLES;
 		}
 
-		sdt_entry = alloc_bootmem(sdt_count * sizeof(struct acpi_table_sdt));
-		if (!sdt_entry) {
-			printk(KERN_ERR "ACPI: Could not allocate mem for SDT entries!\n");
-			return -ENOMEM;
-		}
-
 		for (i = 0; i < sdt_count; i++)
 			sdt_entry[i].pa = (unsigned long) mapped_rsdt->entry[i];
 	}
arch/ia64/kernel/setup.c renamed from arch/ia64/kernel/setup.c.save +1 -1
@@ -378,7 +378,6 @@
 	saved_command_line[COMMAND_LINE_SIZE-1] = '\0';		/* for safety */
 
 	efi_init();
-	find_memory();
 
 #ifdef CONFIG_ACPI_BOOT
 	/* Initialize the ACPI boot-time table parser */
@@ -394,6 +393,7 @@
 #endif /* CONFIG_APCI_BOOT */
 
 	iomem_resource.end = ~0UL;	/* FIXME probably belongs elsewhere */
+	find_memory();
 
 #if 0
 	/* XXX fix me */