OpenBIOS April 2013

openbios@openbios.org

8 participants
89 discussions

[commit] r1130 - trunk/openbios-devel/arch/sparc32

by repository service

Author: mcayland Date: Sat Apr 27 13:30:16 2013 New Revision: 1130 URL: http://tracker.coreboot.org/trac/openbios/changeset/1130 Log: SPARC32: Remove high memory virtual memory reservation The high memory virtual memory reservation introduced in r1114 was a mistake. We need to leave this area as available for clients but simply not allocate from it so that the 0xffc00000 area is free for kadb on older versions of SunOS. Since this part is already handled by the introduction of OFMEM_VIRT_TOP in the same commit then the behaviour is already correct. As reported by Artyom Tarasenko. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/arch/sparc32/openbios.c Modified: trunk/openbios-devel/arch/sparc32/openbios.c ============================================================================== --- trunk/openbios-devel/arch/sparc32/openbios.c Fri Apr 26 11:53:43 2013 (r1129) +++ trunk/openbios-devel/arch/sparc32/openbios.c Sat Apr 27 13:30:16 2013 (r1130) @@ -820,9 +820,6 @@ /* Generate the mapping (and lock translation into the TLBs) */ ofmem_map(phys, virt, MEMORY_SIZE, ofmem_arch_default_translation_mode(phys)); - /* Mark everything up until OFMEM_VIRT_TOP as in use */ - ofmem_claim_virt(OFMEM_VIRT_TOP, OF_CODE_START - MEMORY_SIZE - OFMEM_VIRT_TOP, 0); - /* we push start and end of memory to the stack * so that it can be used by the forth word QUIT * to initialize the memory allocator

8 years

[PATCH 2/2] sparc64: add a eeprom node to the device tree

by Artyom Tarasenko

Add eeprom node to the device tree to make device visible for the guest OS. Signed-off-by: Artyom Tarasenko <atar4qemu(a)gmail.com> --- arch/sparc64/openbios.c | 21 +++++++++++++++++++++ 1 files changed, 21 insertions(+), 0 deletions(-) diff --git a/arch/sparc64/openbios.c b/arch/sparc64/openbios.c index 2e70a9d..22bcc39 100644 --- a/arch/sparc64/openbios.c +++ b/arch/sparc64/openbios.c @@ -450,6 +450,27 @@ void arch_nvram_get(char *data) ob_mmu_init(cpu->name, ram_size); + push_str("/pci@1fe,0/ebus"); + fword("find-device"); + fword("new-device"); + PUSH(0x000001fe); + fword("encode-int"); + PUSH(0x02002000); + fword("encode-int"); + fword("encode+"); + PUSH(NVRAM_SIZE); + fword("encode-int"); + fword("encode+"); + push_str("reg"); + fword("property"); + + push_str("mk48t59"); + fword("model"); + + push_str("eeprom"); + fword("device-name"); + fword("finish-device"); + /* Setup nvram variables */ push_str("/options"); fword("find-device"); -- 1.7.1

8 years

[PATCH 1/2] sparc64: use MMIO for NVRAM access

by Artyom Tarasenko

Use MMIO for accessing the m48t08 NVRAM chip. This patch is a counterpart of a QEMU change. Signed-off-by: Artyom Tarasenko <atar4qemu(a)gmail.com> --- arch/sparc64/openbios.c | 30 ++++++++---------------------- 1 files changed, 8 insertions(+), 22 deletions(-) diff --git a/arch/sparc64/openbios.c b/arch/sparc64/openbios.c index a6c956a..2e70a9d 100644 --- a/arch/sparc64/openbios.c +++ b/arch/sparc64/openbios.c @@ -26,9 +26,7 @@ #define UUID_FMT "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x" -#define NVRAM_ADDR_LO 0x74 -#define NVRAM_ADDR_HI 0x75 -#define NVRAM_DATA 0x77 +#define NVRAM_BASE 0x02000 #define APB_SPECIAL_BASE 0x1fe00000000ULL #define APB_MEM_BASE 0x1ff00000000ULL @@ -40,7 +38,7 @@ #define NVRAM_IDPROM 0x1fd8 #define NVRAM_IDPROM_SIZE 32 #define NVRAM_OB_START (0) -#define NVRAM_OB_SIZE ((0x1fd0 - NVRAM_OB_START) & ~15) +#define NVRAM_OB_SIZE ((NVRAM_IDPROM - NVRAM_OB_START) & ~15) static uint8_t idprom[NVRAM_IDPROM_SIZE]; @@ -333,34 +331,22 @@ id_cpu(void) for (;;); } -static uint8_t nvram_read_byte(uint16_t offset) -{ - outb(offset & 0xff, NVRAM_ADDR_LO); - outb(offset >> 8, NVRAM_ADDR_HI); - return inb(NVRAM_DATA); -} - static void nvram_read(uint16_t offset, char *buf, unsigned int nbytes) { unsigned int i; - for (i = 0; i < nbytes; i++) - buf[i] = nvram_read_byte(offset + i); -} - -static void nvram_write_byte(uint16_t offset, uint8_t val) -{ - outb(offset & 0xff, NVRAM_ADDR_LO); - outb(offset >> 8, NVRAM_ADDR_HI); - outb(val, NVRAM_DATA); + for (i = 0; i < nbytes; i++) { + buf[i] = inb(NVRAM_BASE + offset + i); + } } static void nvram_write(uint16_t offset, const char *buf, unsigned int nbytes) { unsigned int i; - for (i = 0; i < nbytes; i++) - nvram_write_byte(offset + i, buf[i]); + for (i = 0; i < nbytes; i++) { + outb(buf[i], NVRAM_BASE + offset + i); + } } static uint8_t qemu_uuid[16]; -- 1.7.1

8 years

Re: [OpenBIOS] Read errors from SILO booting Debian Wheezy/sparc64

by Mark Cave-Ayland

On 26/04/13 23:12, Artyom Tarasenko wrote: > With the current OpenBIOS (svn.r1129) there are errors reading initrd > from an ext2 partition using SILO. Hi Artyom, Is this a regression? If so, what was the last know good revision? ATB, Mark.

8 years

Re: [OpenBIOS] [commit] r1114 - in trunk/openbios-devel: arch/sparc32 include/arch/sparc32

by Mark Cave-Ayland

On 24/04/13 22:57, Artyom Tarasenko wrote: >> So just to clarify: is what you're saying that a checkout of OpenBIOS r1114 >> fails? In that case can you confirm what value of ofmem_arch_get_virt_top() >> that you were using in the URL above when kadb did work? > > The value (ucell)0xff000000 which you suggested in the SunOS thread, > seems to work > better than the current OFMEM_VIRT_TOP. If I revert the two lines to > the state before r1114: > > - virt = ofmem_claim_virt(OF_CODE_START - MEMORY_SIZE, MEMORY_SIZE, 0); > + virt = ofmem_claim_virt(-1, MEMORY_SIZE, PAGE_SIZE); > > - ofmem_claim_virt(OFMEM_VIRT_TOP, OF_CODE_START - MEMORY_SIZE - > OFMEM_VIRT_TOP, 0); > > then kadb starts with both (ucell)0xff000000 and OFMEM_VIRT_TOP, but > with the later, the kernel dies a bit earlier. Wait a minute - I think this might be a bit of brain-fade on my part. I was thinking that we need to reserve the 0xffc00000 region so that it wouldn't be allocated out to clients, but in actual fact we *do* want clients to use it. We just need to make sure that OpenBIOS itself doesn't return blocks from that region, which is why OFMEM_VIRT_TOP was introduced. In that case can you just try removing this part of the address reservation code from SVN trunk like this: --- a/openbios-devel/arch/sparc32/openbios.c +++ b/openbios-devel/arch/sparc32/openbios.c @@ -820,9 +820,6 @@ static void init_memory(void) /* Generate the mapping (and lock translation into the TLBs) */ ofmem_map(phys, virt, MEMORY_SIZE, ofmem_arch_default_translation_mode(phys)); - /* Mark everything up until OFMEM_VIRT_TOP as in use */ - ofmem_claim_virt(OFMEM_VIRT_TOP, OF_CODE_START - MEMORY_SIZE - OFMEM_VIRT_TOP, 0); - /* we push start and end of memory to the stack * so that it can be used by the forth word QUIT * to initialize the memory allocator Now the only remaining memory reserved in that region is at the very top end, and is just the 128K Forth dictionary which from your tests before should work. ATB, Mark.

8 years

Read errors from SILO booting Debian Wheezy/sparc64

by Artyom Tarasenko

With the current OpenBIOS (svn.r1129) there are errors reading initrd from an ext2 partition using SILO. Steps to reproduce: 1. Install Debian Wheezy RC1 (e.g. as described here http://tyom.blogspot.de/2013/03/debiansparc64-wheezy-under-qemu-how-to.html ) using the standard partition layout ( i.e. /boot parttion with an ext2 file system). 2. Try to boot from the wheezy disk, using it as a -hda image. Below is the boot log with OFMEM_DEBUG enabled. $ sparc64-softmmu/qemu-system-sparc64 -bios /path-to/openbios-devel/obj-sparc64/openbios-builtin.elf -m 384 -nographic -hda /path/debian-wheezy.qcow2 SILO Version 1.4.14 boot: OFMEM: ofmem_map_page_range 0000000040000000 -> 0000000000400000 0000000004000000 mode 0000000000000027 Allocated 64 Megs of memory at 0x40000000 for kernel Uncompressing image... Loaded kernel version 3.2.35 OFMEM: ofmem_map: Bad parameters (0000000004400000 0000000040C00000 000000000094DFFF) ^^^^^^ what happens here? OFMEM: ofmem_map_page_range 0000000040c00000 -> 0000000004400000 000000000094e000 mode 0000000000000027 OFMEM: mapping altered virt=0000000040c00000) Loading initial ramdisk (9747774 bytes at 0x4400000 phys, 0x40C00000 virt)... | Read error on block 1296 (tried 262144, got 4294967295) Read error on block 1296 (tried 262144, got 4294967295) | Read error on block 1810 (tried 262144, got 4294967295) Read error on block 1810 (tried 262144, got 4294967295) \ Read error on block 2324 (tried 242688, got 4294967295) Read error on block 2324 (tried 242688, got 4294967295) | Read error on block 3468 (tried 262144, got 4294967295) Read error on block 3468 (tried 262144, got 4294967295) Read error on block 3468 (tried 262144, got 4294967295) | Read error on block 4239 (tried 262144, got 4294967295) Read error on block 4239 (tried 262144, got 4294967295) \ Read error on block 5780 (tried 1024, got 4294967295) | Read error on block 5524 (tried 262144, got 4294967295) | Read error on block 6295 (tried 262144, got 4294967295) Read error on block 6295 (tried 262144, got 4294967295) OFMEM: ofmem_unmap 0000000040c00000 000000000094dfff [...] [ 17.710187] Unpacking initramfs... [ 17.861207] Initramfs unpacking failed: broken padding [ 17.861323] Freeing initrd memory: 9519k freed -- Regards, Artyom Tarasenko linux/sparc and solaris/sparc under qemu blog: http://tyom.blogspot.com/search/label/qemu

8 years

[commit] r1129 - trunk/openbios-devel/drivers

by repository service

Author: mcayland Date: Fri Apr 26 11:53:43 2013 New Revision: 1129 URL: http://tracker.coreboot.org/trac/openbios/changeset/1129 Log: SPARC: Another escc alignment fix for escc_keyboard Further tests show that the Sun ESCC serial driver routines are also used for the keyboard device. Hence we also need to update the escc_keyboard bindings declaration to use ucell instead of phys_addr_t. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/drivers/escc.c Modified: trunk/openbios-devel/drivers/escc.c ============================================================================== --- trunk/openbios-devel/drivers/escc.c Sun Apr 21 10:27:23 2013 (r1128) +++ trunk/openbios-devel/drivers/escc.c Fri Apr 26 11:53:43 2013 (r1129) @@ -305,7 +305,7 @@ } } -DECLARE_UNNAMED_NODE(escc_keyboard, INSTALL_OPEN, sizeof(phys_addr_t)); +DECLARE_UNNAMED_NODE(escc_keyboard, INSTALL_OPEN, sizeof(ucell)); NODE_METHODS(escc_keyboard) = { { "open", escc_open },

8 years

Re: [OpenBIOS] [commit] r1114 - in trunk/openbios-devel: arch/sparc32 include/arch/sparc32

by Mark Cave-Ayland

On 23/04/13 22:54, Artyom Tarasenko wrote: > Sorry, somehow I missed testing this patch. > It breaks loading kadb: > > 0> boot disk kadb -kdvs Not a bootable ELF image > Loading a.out image... > Loaded 7680 bytes > entry point is 0x4000 > bootpath: /iommu/sbus/espdma/esp/sd@0,0 Ah okay. I thought that the post below was an ack from you that 128K was still okay to run kadb with an unmodified ofmem_arch_get_virt_top(): http://www.openfirmware.info/pipermail/openbios/2013-April/007606.html I also can't see any memory regions in the output above that would overlap 0xffc00000 - has any of the output been removed? So just to clarify: is what you're saying that a checkout of OpenBIOS r1114 fails? In that case can you confirm what value of ofmem_arch_get_virt_top() that you were using in the URL above when kadb did work? ATB, Mark.

8 years

[commit] r1114 - in trunk/openbios-devel: arch/sparc32 include/arch/sparc32

by repository service

Author: mcayland Date: Tue Apr 9 20:32:53 2013 New Revision: 1114 URL: http://tracker.coreboot.org/trac/openbios/changeset/1114 Log: SPARC32: Move Forth dictionary location to a fixed location. Instead of using the automatic allocator to determine the location of the Forth dictionary, allocate it at a fixed location just underneath the ROM itself at 0xffd00000. This patch also prevents corruption within the RAM of the Forth machine by ensuring that we allocate virtual memory outside of OpenBIOS itself, plus reduces the size of the Forth dictionary to 128K so that it meets the memory constraints required to boot Linux and also kadb from older versions of SunOS (as reported by Artyom Tarasenko). Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/arch/sparc32/ofmem_sparc32.c trunk/openbios-devel/arch/sparc32/openbios.c trunk/openbios-devel/include/arch/sparc32/ofmem_sparc32.h Modified: trunk/openbios-devel/arch/sparc32/ofmem_sparc32.c ============================================================================== --- trunk/openbios-devel/arch/sparc32/ofmem_sparc32.c Tue Apr 9 20:32:49 2013 (r1113) +++ trunk/openbios-devel/arch/sparc32/ofmem_sparc32.c Tue Apr 9 20:32:53 2013 (r1114) @@ -61,7 +61,7 @@ ucell ofmem_arch_get_virt_top(void) { - return (ucell)TOP_OF_RAM; + return (ucell)OFMEM_VIRT_TOP; } phys_addr_t ofmem_arch_get_phys_top(void) Modified: trunk/openbios-devel/arch/sparc32/openbios.c ============================================================================== --- trunk/openbios-devel/arch/sparc32/openbios.c Tue Apr 9 20:32:49 2013 (r1113) +++ trunk/openbios-devel/arch/sparc32/openbios.c Tue Apr 9 20:32:53 2013 (r1114) @@ -25,9 +25,9 @@ #include "packages/video.h" #define NO_QEMU_PROTOS #include "arch/common/fw_cfg.h" -#include "libopenbios/ofmem.h" +#include "arch/sparc32/ofmem_sparc32.h" -#define MEMORY_SIZE (512*1024) /* 512K ram for hosted system */ +#define MEMORY_SIZE (128*1024) /* 128K ram for hosted system */ #define UUID_FMT "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x" #define FW_CFG_SUN4M_DEPTH (FW_CFG_ARCH_LOCAL + 0x00) @@ -813,13 +813,16 @@ if (!phys) printk("panic: not enough physical memory on host system.\n"); - virt = ofmem_claim_virt(-1, MEMORY_SIZE, PAGE_SIZE); + virt = ofmem_claim_virt(OF_CODE_START - MEMORY_SIZE, MEMORY_SIZE, 0); if (!virt) printk("panic: not enough virtual memory on host system.\n"); /* Generate the mapping (and lock translation into the TLBs) */ ofmem_map(phys, virt, MEMORY_SIZE, ofmem_arch_default_translation_mode(phys)); + /* Mark everything up until OFMEM_VIRT_TOP as in use */ + ofmem_claim_virt(OFMEM_VIRT_TOP, OF_CODE_START - MEMORY_SIZE - OFMEM_VIRT_TOP, 0); + /* we push start and end of memory to the stack * so that it can be used by the forth word QUIT * to initialize the memory allocator Modified: trunk/openbios-devel/include/arch/sparc32/ofmem_sparc32.h ============================================================================== --- trunk/openbios-devel/include/arch/sparc32/ofmem_sparc32.h Tue Apr 9 20:32:49 2013 (r1113) +++ trunk/openbios-devel/include/arch/sparc32/ofmem_sparc32.h Tue Apr 9 20:32:53 2013 (r1114) @@ -16,6 +16,9 @@ #include "libopenbios/ofmem.h" +#define OF_CODE_START 0xffd00000 +#define OFMEM_VIRT_TOP 0xfe000000 + struct mem; extern struct mem cdvmem;

8 years

[commit] r1128 - trunk/openbios-devel/drivers

by repository service

Author: mcayland Date: Sun Apr 21 10:27:23 2013 New Revision: 1128 URL: http://tracker.coreboot.org/trac/openbios/changeset/1128 Log: SPARC: Fix escc instance structure alignment When binding C functions to Forth words using NODE_METHODs for the SPARC ESCC serial driver, the C instance parameter pointer was set to be a pointer to a phys_addr_t. This is incorrect in the case where sizeof(phys_addr_t) != sizeof(cell) such as on SPARC32, as the address allocated from the Forth machine would be aligned to sizeof(cell); hence an alignment exception could be generated on random occasions on SPARC32 when the allocated address generated by Forth when opening the device was aligned to just 4 bytes (and not also 8 bytes) when dereferencing the instance parameter pointer. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/drivers/escc.c Modified: trunk/openbios-devel/drivers/escc.c ============================================================================== --- trunk/openbios-devel/drivers/escc.c Sun Apr 21 10:27:21 2013 (r1127) +++ trunk/openbios-devel/drivers/escc.c Sun Apr 21 10:27:23 2013 (r1128) @@ -131,7 +131,7 @@ /* ( addr len -- actual ) */ static void -escc_read(phys_addr_t *address) +escc_read(ucell *address) { char *addr; int len; @@ -152,7 +152,7 @@ /* ( addr len -- actual ) */ static void -escc_write(phys_addr_t *address) +escc_write(ucell *address) { unsigned char *addr; int i, len; @@ -172,7 +172,7 @@ } static void -escc_open(phys_addr_t *address) +escc_open(ucell *address) { #ifdef CONFIG_DRIVER_ESCC_SUN int len; @@ -199,7 +199,7 @@ RET ( -1 ); } -DECLARE_UNNAMED_NODE(escc, INSTALL_OPEN, sizeof(phys_addr_t)); +DECLARE_UNNAMED_NODE(escc, INSTALL_OPEN, sizeof(ucell)); NODE_METHODS(escc) = { { "open", escc_open },

8 years

← Newer
1
2
3
4
5
6
7
8
9
Older →