2.4.27, alpha arch, make bootimage and make bootpfile fails

2.4.27, alpha arch, make bootimage and make bootpfile fails

[Norbert Preining]

Hello kernel Gurus!

[debian/woody on alpha sx165/lx164, gcc-3.3.4]

We need to netboot our alphas (because booting from the SCSI controllers
is not an option, they are not supported by the SRM console) and I try
to make a bootp image of our kernel.

When doing this on our alpha the
	make bootimage
and the 
	make bootpfile
both bail out with:
make[1]: Entering directory `/usr/src/linux-2.4.27/arch/alpha/boot'
tools/objstrip -v /usr/src/linux-2.4.27/vmlinux vmlinux.nh
tools/objstrip: extracting 0xfffffc0000310000-0xfffffc00005cfef8 (at
2000)
tools/objstrip: copying 2883320 byte from /usr/src/linux-2.4.27/vmlinux
tools/objstrip: zero-filling bss and aligning to 0 with 393432 bytes
echo "#define KERNEL_SIZE `ls -l vmlinux.nh | awk '{print $5}'`" >
ksize.hT
cmp -s ksize.hT ksize.h || mv -f ksize.hT ksize.h
rm -f ksize.hT
ld -static -T bootloader.lds  head.o main.o
/usr/src/linux-2.4.27/arch/alpha/lib/lib.a
/usr/src/linux-2.4.27/lib/lib.a
/usr/src/linux-2.4.27/arch/alpha/lib/lib.a -o bootloader
/usr/src/linux-2.4.27/lib/lib.a(vsprintf.o): In function `vsnprintf':
vsprintf.o(.text+0xcd4): undefined reference to `printk'
vsprintf.o(.text+0xcdc): undefined reference to `printk'
/usr/src/linux-2.4.27/lib/lib.a(dump_stack.o): In function `dump_stack':
dump_stack.o(.text+0x10): undefined reference to `printk'
dump_stack.o(.text+0x1c): undefined reference to `printk'
make[1]: *** [bootloader] Error 1
make[1]: Leaving directory `/usr/src/linux-2.4.27/arch/alpha/boot'
make: *** [bootimage] Error 2

and similar with bootpfile

Is there a way around this, we definitely need the bootp image.

Thanks a lot and all the best

Norbert

-------------------------------------------------------------------------------
Norbert Preining <preining AT logic DOT at>         Technische Universität Wien
gpg DSA: 0x09C5B094      fp: 14DF 2E6C 0307 BE6D AD76  A9C0 D2BF 4AA3 09C5 B094
-------------------------------------------------------------------------------
SCREEB (n.)
To make the noise of a nylon anorak rubbing against a pair of corduroy
trousers.
			--- Douglas Adams, The Meaning of Liff

Re: 2.4.27, alpha arch, make bootimage and make bootpfile fails

[Ivan Kokshaysky]

On Tue, Oct 12, 2004 at 07:33:44PM +0200, Norbert Preining wrote:
> When doing this on our alpha the
> 	make bootimage
> and the 
> 	make bootpfile
> both bail out with:
...
> /usr/src/linux-2.4.27/arch/alpha/lib/lib.a -o bootloader
> /usr/src/linux-2.4.27/lib/lib.a(vsprintf.o): In function `vsnprintf':
> vsprintf.o(.text+0xcd4): undefined reference to `printk'

Thanks for the report. The appended patch should fix that.

Ivan.

--- 2.4/arch/alpha/boot/bootloader.lds	Tue Jul  3 01:40:14 2001
+++ linux/arch/alpha/boot/bootloader.lds	Wed Oct 13 23:18:50 2004
@@ -1,5 +1,6 @@
 OUTPUT_FORMAT("elf64-alpha")
 ENTRY(__start)
+printk = srm_printk;
 SECTIONS
 {
   . = 0x20000000;

Re: 2.4.27, alpha arch, make bootimage and make bootpfile fails

[Norbert Preining]

On Mit, 13 Okt 2004, Ivan Kokshaysky wrote:
> > /usr/src/linux-2.4.27/arch/alpha/lib/lib.a -o bootloader
> > /usr/src/linux-2.4.27/lib/lib.a(vsprintf.o): In function `vsnprintf':
> > vsprintf.o(.text+0xcd4): undefined reference to `printk'
> 
> Thanks for the report. The appended patch should fix that.

Thanks a lot. This fixed it. Testing will be done tomorrow.

Best wishes

Norbert

-------------------------------------------------------------------------------
Norbert Preining <preining AT logic DOT at>         Technische Universität Wien
gpg DSA: 0x09C5B094      fp: 14DF 2E6C 0307 BE6D AD76  A9C0 D2BF 4AA3 09C5 B094
-------------------------------------------------------------------------------
WHAPLODE DROVE (n.)
A homicidal golf stroke.
			--- Douglas Adams, The Meaning of Liff

Re: 2.4.27, alpha arch, make bootimage and make bootpfile fails

[Norbert Preining]

[-- Attachment #1: Type: text/plain, Size: 1069 bytes --]

Dear Ivan, dear kernel kackers!

On Mit, 13 Okt 2004, Ivan Kokshaysky wrote:
> > 	make bootpfile

Ok, now I have a boot file but the kernel is not started:

>>> boot ewa0 -fl ...
...
jumping to bootstrap code
Linux/AXP bootp loader for linux 2.4.27
Switching to OSF PAL-code .. 0k (rev 1000800020117)
Loading the kernel...'root=/dev/sda1'

Halted CPU 0

halt code = 2
kernel stack not valid halt
PC = 200000000
boot failure
>>>

(typed in from screen)

SRM 5.5, AlphaPC164
normal (from scsi) boot dmesg with aboot is attached.

Any hints for this?

Best wishes

Norbert

-------------------------------------------------------------------------------
Norbert Preining <preining AT logic DOT at>         Technische Universität Wien
gpg DSA: 0x09C5B094      fp: 14DF 2E6C 0307 BE6D AD76  A9C0 D2BF 4AA3 09C5 B094
-------------------------------------------------------------------------------
HARPENDEN (n.)
The coda to a phone conversion, consisting of about eight exchanges,
by which people try gracefully to get off the line.
			--- Douglas Adams, The Meaning of Liff

[-- Attachment #2: alpha.dmesg --]
[-- Type: text/plain, Size: 3657 bytes --]

Linux version 2.4.27alphafullscsi (root@beta) (gcc version 3.3.4) #1 Tue Oct 12 08:04:47 CEST 2004
Booting on EB164 variation PC164 using machine vector PC164 from SRM
Major Options: EV56 LEGACY_START MAGIC_SYSRQ 
Command line: ro root=/dev/sda1
memcluster 0, usage 1, start        0, end      192
memcluster 1, usage 0, start      192, end    32651
memcluster 2, usage 1, start    32651, end    32768
freeing pages 192:384
freeing pages 794:32651
reserving pages 794:795
pci: cia revision 2
On node 0 totalpages: 32651
zone(0): 32651 pages.
zone(1): 0 pages.
zone(2): 0 pages.
Kernel command line: ro root=/dev/sda1
Using epoch = 2000
Console: colour VGA+ 80x25
Calibrating delay loop... 994.44 BogoMIPS
Memory: 253544k/261208k available (2070k kernel code, 6128k reserved, 351k data, 328k init)
Dentry cache hash table entries: 32768 (order: 6, 524288 bytes)
Inode cache hash table entries: 16384 (order: 5, 262144 bytes)
Mount cache hash table entries: 512 (order: 0, 8192 bytes)
Buffer cache hash table entries: 8192 (order: 3, 65536 bytes)
Page-cache hash table entries: 32768 (order: 5, 262144 bytes)
POSIX conformance testing by UNIFIX
pci: passed tb register update test
pci: passed sg loopback i/o read test
pci: passed tbia test
pci: passed pte write cache snoop test
pci: failed valid tag invalid pte reload test (mcheck; workaround available)
pci: passed pci machine check test
pci: enabling save/restore of SRM state
SMC FDC37C93X Ultra I/O Controller found @ 0x370
Linux NET4.0 for Linux 2.4
Based upon Swansea University Computer Society NET3.039
Initializing RT netlink socket
srm_env: version 0.0.5 loaded successfully
Starting kswapd
VFS: Disk quotas vdquot_6.5.1
Journalled Block Device driver loaded
Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
pty: 256 Unix98 ptys configured
Serial driver version 5.05c (2001-07-08) with MANY_PORTS SHARE_IRQ SERIAL_PCI enabled
ttyS00 at 0x03f8 (irq = 4) is a 16550A
ttyS01 at 0x02f8 (irq = 3) is a 16550A
rtc: SRM (post-2000) epoch (2000) detected
Real Time Clock Driver v1.10f
Floppy drive(s): fd0 is 2.88M
FDC 0 is a post-1991 82077
eth0: DE434/5 at 0x8400 (PCI bus 0, device 7), h/w address 00:c0:95:ec:b7:60,
      and requires IRQ17 (provided by PCI BIOS).
de4x5.c:V0.546 2001/02/22 davies@maniac.ultranet.com
SCSI subsystem driver Revision: 1.00
sym0: <810> rev 0x2 on pci bus 0 device 9 function 0 irq 19
sym0: No NVRAM, ID 7, Fast-10, SE, parity checking
sym0: SCSI BUS has been reset.
scsi0 : sym-2.1.17a
blk: queue fffffc000fe2eca8, no I/O memory limit
  Vendor: IBM       Model: DNES-309170       Rev: SA30
  Type:   Direct-Access                      ANSI SCSI revision: 03
blk: queue fffffc000fe2eea8, no I/O memory limit
  Vendor: HP        Model: C1533A            Rev: 9503
  Type:   Sequential-Access                  ANSI SCSI revision: 02
blk: queue fffffc000fe2f0a8, no I/O memory limit
sym0:0:0: tagged command queuing enabled, command queue depth 16.
st: Version 20040102, bufsize 32768, max init. bufs 4, s/g segs 16
Attached scsi tape st0 at scsi0, channel 0, id 5, lun 0
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
sym0:0: FAST-10 SCSI 10.0 MB/s ST (100.0 ns, offset 8)
SCSI device sda: 17916240 512-byte hdwr sectors (9173 MB)
Partition check:
 sda: sda1 sda2 sda3 sda4
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP
IP: routing cache hash table of 4096 buckets, 32Kbytes
TCP: Hash tables configured (established 32768 bind 32768)
NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
VFS: Mounted root (ext2 filesystem) readonly.
Freeing unused kernel memory: 328k freed
Adding Swap: 205592k swap-space (priority -1)
eth0: media is TP.

Re: 2.4.27, alpha arch, make bootimage and make bootpfile fails

[Måns Rullgård]

Norbert Preining <preining@logic.at> writes:

> Dear Ivan, dear kernel kackers!
>
> On Mit, 13 Okt 2004, Ivan Kokshaysky wrote:
>> > 	make bootpfile
>
> Ok, now I have a boot file but the kernel is not started:
>
>>>> boot ewa0 -fl ...
> ...
> jumping to bootstrap code
> Linux/AXP bootp loader for linux 2.4.27
> Switching to OSF PAL-code .. 0k (rev 1000800020117)
> Loading the kernel...'root=/dev/sda1'
>
> Halted CPU 0
>
> halt code = 2
> kernel stack not valid halt
> PC = 200000000
> boot failure

Try building more things as modules instead of builtin, if possible.

-- 
Måns Rullgård
mru@mru.ath.cx

Re: 2.4.27, alpha arch, make bootimage and make bootpfile fails

[Norbert Preining]

On Don, 14 Okt 2004, Måns Rullgård wrote:
> > Loading the kernel...'root=/dev/sda1'
> >
> > Halted CPU 0
> >
> > halt code = 2
> > kernel stack not valid halt
> > PC = 200000000
> > boot failure
> 
> Try building more things as modules instead of builtin, if possible.

I have reduced the whole kernel zo more or less nothing included,
besides the SCSI drivers, and the same still happens. 

Best wishes

Norbert

-------------------------------------------------------------------------------
Norbert Preining <preining AT logic DOT at>         Technische Universität Wien
gpg DSA: 0x09C5B094      fp: 14DF 2E6C 0307 BE6D AD76  A9C0 D2BF 4AA3 09C5 B094
-------------------------------------------------------------------------------
LOCHRANZA (n.)
The long unaccomplished wail in the middle of a Scottish folk song
where the pipes nip around the corner for a couple of drinks.
			--- Douglas Adams, The Meaning of Liff

Re: 2.4.27, alpha arch, make bootimage and make bootpfile fails

[Ivan Kokshaysky]

On Thu, Oct 14, 2004 at 03:00:35PM +0200, Norbert Preining wrote:
> Ok, now I have a boot file but the kernel is not started:
> 
> >>> boot ewa0 -fl ...
> ...
> Halted CPU 0
> 
> halt code = 2
> kernel stack not valid halt

I was able to reproduce that on SX164 with both bootpfile and bootpzfile.
As it turns out, the final kernel destination overlaps the initial stack
region. Please test the appended patch.

Ivan.

--- 2.4/arch/alpha/boot/bootp.c	Fri Oct  5 05:47:08 2001
+++ linux/arch/alpha/boot/bootp.c	Tue Oct 19 19:22:21 2004
@@ -26,6 +26,8 @@ extern unsigned long switch_to_osf_pal(u
 	struct pcb_struct * pcb_va, struct pcb_struct * pcb_pa,
 	unsigned long *vptb);
 
+extern void move_stack(unsigned long new_stack);
+
 struct hwrpb_struct *hwrpb = INIT_HWRPB;
 static struct pcb_struct pcb_va[1];
 
@@ -118,12 +120,10 @@ static inline void
 runkernel(void)
 {
 	__asm__ __volatile__(
-		"bis %1,%1,$30\n\t"
 		"bis %0,%0,$27\n\t"
 		"jmp ($27)"
 		: /* no outputs: it doesn't even return */
-		: "r" (START_ADDR),
-		  "r" (PAGE_SIZE + INIT_STACK));
+		: "r" (START_ADDR));
 }
 
 extern char _end;
@@ -147,9 +147,7 @@ start_kernel(void)
 	 */
 	static long nbytes;
 	static char envval[256] __attribute__((aligned(8)));
-#ifdef INITRD_IMAGE_SIZE
 	static unsigned long initrd_start;
-#endif
 
 	srm_printk("Linux/AXP bootp loader for Linux " UTS_RELEASE "\n");
 	if (INIT_HWRPB->pagesize != 8192) {
@@ -164,12 +162,19 @@ start_kernel(void)
 	}
 	pal_init();
 
-#ifdef INITRD_IMAGE_SIZE
 	/* The initrd must be page-aligned.  See below for the 
 	   cause of the magic number 5.  */
-	initrd_start = ((START_ADDR + 5*KERNEL_SIZE) | (PAGE_SIZE-1)) + 1;
+	initrd_start = ((START_ADDR + 5*KERNEL_SIZE + PAGE_SIZE) |
+			(PAGE_SIZE-1)) + 1;
+#ifdef INITRD_IMAGE_SIZE
 	srm_printk("Initrd positioned at %#lx\n", initrd_start);
 #endif
+
+	/*
+	 * Move the stack to a safe place to ensure it won't be
+	 * overwritten by kernel image.
+	 */
+	move_stack(initrd_start - PAGE_SIZE);
 
 	nbytes = callback_getenv(ENV_BOOTED_OSFLAGS, envval, sizeof(envval));
 	if (nbytes < 0 || nbytes >= sizeof(envval)) {
--- 2.4/arch/alpha/boot/bootpz.c	Tue Oct 19 19:27:49 2004
+++ linux/arch/alpha/boot/bootpz.c	Tue Oct 19 19:12:16 2004
@@ -32,9 +32,6 @@
 #undef DEBUG_CHECK_RANGE
 #define DEBUG_ADDRESSES
 
-#define DEBUG_SP(x) \
-    {register long sp asm("30"); srm_printk("%s (sp=%lx)\n", x, sp);}
-
 extern unsigned long switch_to_osf_pal(unsigned long nr,
 	struct pcb_struct * pcb_va, struct pcb_struct * pcb_pa,
 	unsigned long *vptb);
@@ -42,6 +39,8 @@ extern unsigned long switch_to_osf_pal(u
 extern int decompress_kernel(void* destination, void *source,
 			     size_t ksize, size_t kzsize);
 
+extern void move_stack(unsigned long new_stack);
+
 struct hwrpb_struct *hwrpb = INIT_HWRPB;
 static struct pcb_struct pcb_va[1];
 
@@ -154,12 +153,10 @@ static inline void
 runkernel(void)
 {
 	__asm__ __volatile__(
-		"bis %1,%1,$30\n\t"
 		"bis %0,%0,$27\n\t"
 		"jmp ($27)"
 		: /* no outputs: it doesn't even return */
-		: "r" (START_ADDR),
-		  "r" (PAGE_SIZE + INIT_STACK));
+		: "r" (START_ADDR));
 }
 
 /* Must record the SP (it is virtual) on entry, so we can make sure
@@ -244,7 +241,9 @@ extern char _end;
    for "bootmem" anyway.
 */
 #define K_COPY_IMAGE_START	NEXT_PAGE(K_KERNEL_IMAGE_END)
-#define K_INITRD_START		NEXT_PAGE(K_COPY_IMAGE_START + KERNEL_SIZE)
+/* Reserve one page below INITRD for the new stack. */
+#define K_INITRD_START \
+    NEXT_PAGE(K_COPY_IMAGE_START + KERNEL_SIZE + PAGE_SIZE)
 #define K_COPY_IMAGE_END \
     (K_INITRD_START + REAL_INITRD_SIZE + MALLOC_AREA_SIZE)
 #define K_COPY_IMAGE_SIZE \
@@ -410,8 +409,7 @@ start_kernel(void)
 		   initrd_image_start,
 		   INITRD_IMAGE_SIZE);
 #endif
-	memcpy((void *)initrd_image_start,
-	       (void *)V_INITRD_START,
+	memcpy((void *)initrd_image_start, (void *)V_INITRD_START,
 	       INITRD_IMAGE_SIZE);
 
 #endif /* INITRD_IMAGE_SIZE */
@@ -427,9 +425,14 @@ start_kernel(void)
 			   K_KERNEL_IMAGE_START,
 			   (unsigned)KERNEL_SIZE);
 #endif
+		/*
+		 * Move the stack to a safe place to ensure it won't be
+		 * overwritten by kernel image.
+		 */
+		move_stack(initrd_image_start - PAGE_SIZE);
+
 		memcpy((void *)K_KERNEL_IMAGE_START,
-		       (void *)uncompressed_image_start,
-		       KERNEL_SIZE);
+		       (void *)uncompressed_image_start, KERNEL_SIZE);
 	}
 	
 	/* Clear the zero page, then move the argument list in. */
--- 2.4/arch/alpha/boot/bootloader.lds	Tue Jul  3 01:40:14 2001
+++ linux/arch/alpha/boot/bootloader.lds	Tue Oct 19 15:56:21 2004
@@ -1,5 +1,6 @@
 OUTPUT_FORMAT("elf64-alpha")
 ENTRY(__start)
+printk = srm_printk;
 SECTIONS
 {
   . = 0x20000000;
--- 2.4/arch/alpha/boot/head.S	Mon Oct 12 22:40:12 1998
+++ linux/arch/alpha/boot/head.S	Tue Oct 19 18:51:47 2004
@@ -100,3 +100,24 @@ halt:
 	.prologue 0
 	call_pal PAL_halt
 	.end halt
+
+/* $16 - new stack page */
+	.align 3
+	.globl	move_stack
+	.ent	move_stack
+move_stack:
+	.prologue 0
+	lda	$0, 0x1fff($31)
+	and	$0, $30, $1			/* Stack offset */
+	or	$1, $16, $16			/* New stack pointer */
+	mov	$30, $1
+	mov	$16, $2
+1:	ldq	$3, 0($1)			/* Move the stack */
+	addq	$1, 8, $1
+	stq	$3, 0($2)
+	and	$0, $1, $4
+	addq	$2, 8, $2
+	bne	$4, 1b
+	mov	$16, $30
+	ret	($26)
+	.end move_stack

Re: 2.4.27, alpha arch, make bootimage and make bootpfile fails

[Norbert Preining]

On Die, 19 Okt 2004, Ivan Kokshaysky wrote:
> I was able to reproduce that on SX164 with both bootpfile and bootpzfile.
> As it turns out, the final kernel destination overlaps the initial stack
> region. Please test the appended patch.

Bingo. Thanks a lot. I tested bootpfile and this worked without any
problems. Very nice.

Best wishes

Norbert

-------------------------------------------------------------------------------
Norbert Preining <preining AT logic DOT at>         Technische Universität Wien
gpg DSA: 0x09C5B094      fp: 14DF 2E6C 0307 BE6D AD76  A9C0 D2BF 4AA3 09C5 B094
-------------------------------------------------------------------------------
TUMBY (n.)
The involuntary abdominal gurgling which fills the silence following
someone else's intimate personal revelation.
			--- Douglas Adams, The Meaning of Liff