OpenBIOS November 2010

openbios@openbios.org

10 participants
59 discussions

a couple of problem while compiling and running openfirmware on qemu
by Gleb Natapov 13 Nov '10

13 Nov '10

[Resending. Wasn't subscribed to the list and it looks like it is required to post, but if first mail got through sorry] Hello, I tried to compile openfirmware for qemu like described here http://www.openfirmware.info/Building_OFW_for_QEMU. But after running make in cpu/x86/pc/emu/build on 64bit host I get linkage error: ld -T inflate.ld inflate.lo -o inflate.o ld: i386 architecture of input file `inflate.lo' is incompatible with i386:x86-64 output After applying this patch: Index: cpu/x86/Linux/Makefile =================================================================== --- cpu/x86/Linux/Makefile (revision 1996) +++ cpu/x86/Linux/Makefile (working copy) @@ -34,7 +34,7 @@ ${CC} -c -m32 -Wall -fno-stack-protector -ffreestanding -D_FORTIFY_SOURCE=0 -DNEED_BCOPY -O3 -fpic $< -o $@ inflate.o: inflate.lo - ${LD} -T inflate.ld $< -o $@ + ${LD} -melf_i386 -T inflate.ld $< -o $@ ../build/inflate.bin: inflate.o objcopy -O binary $< $@ ==== compilation goes a little bit further but now it stops with: ./build emuofw.rom --- Rebuilding ohci.fc --- Cmd: ${HOSTDIR}/forth ${HOSTDIR}/../build/builder.dic ../ohci.bth ${BP}/dev/usb2/hcd/ohci/ohci.fth:25: my-w@ ? don't know much forth so just commented every line that gave me an error: Index: dev/usb2/hcd/ohci/ohci.fth =================================================================== --- dev/usb2/hcd/ohci/ohci.fth (revision 1996) +++ dev/usb2/hcd/ohci/ohci.fth (working copy) @@ -22,12 +22,12 @@ 0 value ohci-reg : map-regs ( -- ) - 4 my-w@ h# 16 or 4 my-w! - 0 0 my-space h# 0200.0010 + 1000 map-in to ohci-reg +\ 4 my-w@ h# 16 or 4 my-w! +\ 0 0 my-space h# 0200.0010 + 1000 map-in to ohci-reg ; : unmap-regs ( -- ) - ohci-reg 1000 map-out 0 to ohci-reg +\ ohci-reg 1000 map-out 0 to ohci-reg ; : ohci-reg@ ( idx -- data ) ohci-reg + rl@ ; ======== After that compilation finally succeeded. Produced rom image runs fine on qemu, but fails on qemu-kvm with error: BUG: kvm_dirty_pages_log_change: invalid parameters 0000000010000000-0000000010ffffff Looking into it it seams like openfirmware configure cirrus framebuffer to be at address 0x10000000, but later it configures nic memory BAR with the same address (during device enumeration perhaps). Later it deconfigures nic again. Qemu somehow handles this fine, but qemu-kvm doesn't. This shouldn't work on real HW too FWIW. -- Gleb.

3 3

[commit] r952 - trunk/openbios-devel/drivers
by repository service 10 Nov '10

10 Nov '10

Author: afaerber Date: Wed Nov 10 23:35:39 2010 New Revision: 952 URL: http://tracker.coreboot.org/trac/openbios/changeset/952 Log: cuda: Use FMT_plx Use new FMT_plx format for ppc64-compatible trace output. Signed-off-by: Andreas Färber <andreas.faerber(a)web.de> Modified: trunk/openbios-devel/drivers/cuda.c Modified: trunk/openbios-devel/drivers/cuda.c ============================================================================== --- trunk/openbios-devel/drivers/cuda.c Tue Nov 9 23:50:48 2010 (r951) +++ trunk/openbios-devel/drivers/cuda.c Wed Nov 10 23:35:39 2010 (r952) @@ -376,7 +376,7 @@ phandle_t aliases; base += IO_CUDA_OFFSET; - CUDA_DPRINTF(" base=%08x\n", base); + CUDA_DPRINTF(" base=" FMT_plx "\n", base); cuda = malloc(sizeof(cuda_t)); if (cuda == NULL) return NULL;

1 0

[commit] r951 - in trunk/openbios-devel: include/kernel kernel
by repository service 09 Nov '10

09 Nov '10

Author: afaerber Date: Tue Nov 9 23:50:48 2010 New Revision: 951 URL: http://tracker.coreboot.org/trac/openbios/changeset/951 Log: Convert cell2pointer() and pointer2cell() macros to inline functions Inline functions suggested by Blue. Use const void* for pointer2cell() due to const char* arguments. Fix a misuse of pointer2cell(). Drop comment on reusing different host bitness code path; any special handling for different target bitnesses should go into these new functions. v2: * Use QEMU-style indentation for new inline functions. Cc: Blue Swirl <blauwirbel(a)gmail.com> Signed-off-by: Andreas Färber <andreas.faerber(a)web.de> Modified: trunk/openbios-devel/include/kernel/stack.h trunk/openbios-devel/kernel/internal.c Modified: trunk/openbios-devel/include/kernel/stack.h ============================================================================== --- trunk/openbios-devel/include/kernel/stack.h Tue Nov 9 21:12:35 2010 (r950) +++ trunk/openbios-devel/include/kernel/stack.h Tue Nov 9 23:50:48 2010 (r951) @@ -31,9 +31,17 @@ #ifdef NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH -// XXX Check whether we can take the larger-than code path for ppc64 instead. -#define pointer2cell(x) ((ucell)(uintptr_t)(x)) -#define cell2pointer(x) ((u8 *)(uintptr_t)(x)) + +static inline ucell pointer2cell(const void* x) +{ + return (ucell)(uintptr_t)x; +} + +static inline void* cell2pointer(ucell x) +{ + return (void*)(uintptr_t)x; +} + #endif static inline void PUSH(ucell value) { Modified: trunk/openbios-devel/kernel/internal.c ============================================================================== --- trunk/openbios-devel/kernel/internal.c Tue Nov 9 21:12:35 2010 (r950) +++ trunk/openbios-devel/kernel/internal.c Tue Nov 9 23:50:48 2010 (r951) @@ -645,7 +645,7 @@ case 'F': /* Start subordinate Forth interpreter */ PUSHR(PC - sizeof(cell)); - PC = pointer2cell(findword("outer-interpreter")) + sizeof(ucell); + PC = findword("outer-interpreter") + sizeof(ucell); /* Save rstack position for when we return */ dbgrstackcnt = rstackcnt;

1 0

[commit] r950 - trunk/openbios-devel/config/scripts
by repository service 09 Nov '10

09 Nov '10

Author: blueswirl Date: Tue Nov 9 21:12:35 2010 New Revision: 950 URL: http://tracker.coreboot.org/trac/openbios/changeset/950 Log: Add support to -linux-gcc style cross compilers Check for -linux-gcc cross compilers in addition to -linux-gnu-gcc. Signed-off-by: Blue Swirl <blauwirbel(a)gmail.com> Modified: trunk/openbios-devel/config/scripts/switch-arch Modified: trunk/openbios-devel/config/scripts/switch-arch ============================================================================== --- trunk/openbios-devel/config/scripts/switch-arch Mon Nov 8 22:12:36 2010 (r949) +++ trunk/openbios-devel/config/scripts/switch-arch Tue Nov 9 21:12:35 2010 (r950) @@ -97,7 +97,7 @@ select_prefix() { - for TARGET in ${1}-linux-gnu- ${1}-elf- ${1}-eabi- + for TARGET in ${1}-linux-gnu- ${1}-linux- ${1}-elf- ${1}-eabi- do if type ${TARGET}gcc > /dev/null 2>&1 then

1 0

[PATCH] Introduce forth_init() for trampoline initialization
by Andreas Färber 09 Nov '10

09 Nov '10

Move trampoline variable initialization to init_trampoline(), like in kernel/bootstrap.c. Add calls to a new forth_init() function for each architecture to invoke it. Idea courtesy of Blue. This fixes ppc64 compilation by avoiding a casted self-reference. Cc: Blue Swirl <blauwirbel(a)gmail.com> Signed-off-by: Andreas Färber <andreas.faerber(a)web.de> --- Tested this on ppc, sparc32, sparc64 - OpenBIOS still boots up. forth_init() placement in internal.c may not be perfect but it fits its current purpose... arch/amd64/openbios.c | 1 + arch/ppc/kernel.c | 1 + arch/ppc/qemu/kernel.c | 1 + arch/sparc32/openbios.c | 1 + arch/sparc64/openbios.c | 1 + arch/unix/unix.c | 1 + arch/x86/openbios.c | 1 + include/kernel/kernel.h | 1 + kernel/internal.c | 15 ++++++++++++++- 9 files changed, 22 insertions(+), 1 deletions(-) diff --git a/arch/amd64/openbios.c b/arch/amd64/openbios.c index df43943..db138ad 100644 --- a/arch/amd64/openbios.c +++ b/arch/amd64/openbios.c @@ -67,6 +67,7 @@ int openbios(void) load_dictionary((char *)sys_info.dict_start, sys_info.dict_end-sys_info.dict_start); + forth_init(); relocate(&sys_info); diff --git a/arch/ppc/kernel.c b/arch/ppc/kernel.c index 57efde7..28f2965 100644 --- a/arch/ppc/kernel.c +++ b/arch/ppc/kernel.c @@ -83,6 +83,7 @@ initialize_forth( void ) { dict = malloc(DICTIONARY_SIZE); load_dictionary( forth_dictionary, sizeof(forth_dictionary) ); + forth_init(); PUSH_xt( bind_noname_func(arch_of_init) ); fword("PREPOST-initializer"); diff --git a/arch/ppc/qemu/kernel.c b/arch/ppc/qemu/kernel.c index dbfca57..4cae525 100644 --- a/arch/ppc/qemu/kernel.c +++ b/arch/ppc/qemu/kernel.c @@ -86,6 +86,7 @@ initialize_forth( void ) dictlimit = DICTIONARY_SIZE; load_dictionary( forth_dictionary, sizeof(forth_dictionary) ); + forth_init(); PUSH_xt( bind_noname_func(arch_of_init) ); fword("PREPOST-initializer"); diff --git a/arch/sparc32/openbios.c b/arch/sparc32/openbios.c index e368d6c..1c86752 100644 --- a/arch/sparc32/openbios.c +++ b/arch/sparc32/openbios.c @@ -970,6 +970,7 @@ int openbios(void) load_dictionary((char *)sys_info.dict_start, (unsigned long)sys_info.dict_end - (unsigned long)sys_info.dict_start); + forth_init(); #ifdef CONFIG_DEBUG_BOOT printk("forth started.\n"); diff --git a/arch/sparc64/openbios.c b/arch/sparc64/openbios.c index 49bbe27..8c08814 100644 --- a/arch/sparc64/openbios.c +++ b/arch/sparc64/openbios.c @@ -623,6 +623,7 @@ int openbios(void) load_dictionary((char *)sys_info.dict_start, (unsigned long)sys_info.dict_end - (unsigned long)sys_info.dict_start); + forth_init(); #ifdef CONFIG_DEBUG_BOOT printk("forth started.\n"); diff --git a/arch/unix/unix.c b/arch/unix/unix.c index f1c18df..5e95b69 100644 --- a/arch/unix/unix.c +++ b/arch/unix/unix.c @@ -516,6 +516,7 @@ int main(int argc, char *argv[]) printk("done.\n"); read_dictionary(argv[optind]); + forth_init(); PUSH_xt( bind_noname_func(arch_init) ); fword("PREPOST-initializer"); diff --git a/arch/x86/openbios.c b/arch/x86/openbios.c index 62ef587..24b886d 100644 --- a/arch/x86/openbios.c +++ b/arch/x86/openbios.c @@ -90,6 +90,7 @@ int openbios(void) load_dictionary((char *)sys_info.dict_start, (unsigned long)sys_info.dict_end - (unsigned long)sys_info.dict_start); + forth_init(); relocate(&sys_info); diff --git a/include/kernel/kernel.h b/include/kernel/kernel.h index 15605b5..87812e9 100644 --- a/include/kernel/kernel.h +++ b/include/kernel/kernel.h @@ -32,6 +32,7 @@ extern void panic(const char *error) __attribute__ ((noreturn)); extern xt_t findword(const char *s1); extern void modules_init( void ); +extern void forth_init(void); /* arch kernel hooks */ extern void exception(cell no); diff --git a/kernel/internal.c b/kernel/internal.c index 760c66d..0e3794a 100644 --- a/kernel/internal.c +++ b/kernel/internal.c @@ -50,8 +50,21 @@ char xtname[MAXNFALEN]; /* instead of pointing to an explicit 0 variable we * point behind the pointer. */ -static ucell t[] = { DOCOL, 0, (ucell)(t+3), 0 }; +static ucell t[] = { 0, 0, 0, 0 }; static ucell *trampoline = t; + +static void init_trampoline(void) +{ + trampoline[0] = DOCOL; + trampoline[1] = 0; + trampoline[2] = pointer2cell(&trampoline[3]); + trampoline[3] = 0; +} + +void forth_init(void) +{ + init_trampoline(); +} #endif #ifndef CONFIG_DEBUG_INTERPRETER -- 1.7.3

2 2

[commit] r949 - in trunk/openbios-devel/include/arch: amd64 ia64 ppc sparc32 sparc64 x86
by repository service 09 Nov '10

09 Nov '10

Author: afaerber Date: Mon Nov 8 22:12:36 2010 New Revision: 949 URL: http://tracker.coreboot.org/trac/openbios/changeset/949 Log: Introduce FMT_plx for phys_addr_t Define a zero-padded format string to be used for phys_addr_t arguments. Introduce PRIx{32,64} macros as needed. v2: * Also define PRIx32 for sparc32 and sparc64, requested by Blue. Cc: Mark Cave-Ayland <mark.cave-ayland(a)siriusit.co.uk> Cc: Blue Swirl <blauwirbel(a)gmail.com> Signed-off-by: Andreas Färber <andreas.faerber(a)web.de> Modified: trunk/openbios-devel/include/arch/amd64/types.h trunk/openbios-devel/include/arch/ia64/types.h trunk/openbios-devel/include/arch/ppc/types.h trunk/openbios-devel/include/arch/sparc32/types.h trunk/openbios-devel/include/arch/sparc64/types.h trunk/openbios-devel/include/arch/x86/types.h Modified: trunk/openbios-devel/include/arch/amd64/types.h ============================================================================== --- trunk/openbios-devel/include/arch/amd64/types.h Sun Nov 7 19:08:41 2010 (r948) +++ trunk/openbios-devel/include/arch/amd64/types.h Mon Nov 8 22:12:36 2010 (r949) @@ -17,8 +17,11 @@ #include "autoconf.h" /* physical address */ + typedef uint64_t phys_addr_t; +#define FMT_plx "%016" PRIx64 + /* cell based types */ typedef long long cell; Modified: trunk/openbios-devel/include/arch/ia64/types.h ============================================================================== --- trunk/openbios-devel/include/arch/ia64/types.h Sun Nov 7 19:08:41 2010 (r948) +++ trunk/openbios-devel/include/arch/ia64/types.h Mon Nov 8 22:12:36 2010 (r949) @@ -18,8 +18,11 @@ #include <endian.h> /* physical address */ + typedef uint64_t phys_addr_t; +#define FMT_plx "%016" PRIx64 + /* cell based types */ typedef int64_t cell; Modified: trunk/openbios-devel/include/arch/ppc/types.h ============================================================================== --- trunk/openbios-devel/include/arch/ppc/types.h Sun Nov 7 19:08:41 2010 (r948) +++ trunk/openbios-devel/include/arch/ppc/types.h Mon Nov 8 22:12:36 2010 (r949) @@ -25,6 +25,9 @@ typedef int int32_t; typedef long long int64_t; typedef long intptr_t; + +#define PRIx32 "x" +#define PRIx64 "llx" #endif /* endianess */ @@ -33,8 +36,10 @@ /* physical address */ #if defined(__powerpc64__) typedef uint64_t phys_addr_t; +#define FMT_plx "%016" PRIx64 #else typedef uint32_t phys_addr_t; +#define FMT_plx "%08" PRIx32 #endif /* cell based types */ Modified: trunk/openbios-devel/include/arch/sparc32/types.h ============================================================================== --- trunk/openbios-devel/include/arch/sparc32/types.h Sun Nov 7 19:08:41 2010 (r948) +++ trunk/openbios-devel/include/arch/sparc32/types.h Mon Nov 8 22:12:36 2010 (r949) @@ -25,14 +25,20 @@ typedef int int32_t; typedef long long int64_t; typedef long intptr_t; + +#define PRIx32 "x" +#define PRIx64 "llx" #endif /* endianess */ #include "autoconf.h" /* physical address: 36 bits */ + typedef uint64_t phys_addr_t; +#define FMT_plx "%09" PRIx64 + /* cell based types */ typedef int32_t cell; Modified: trunk/openbios-devel/include/arch/sparc64/types.h ============================================================================== --- trunk/openbios-devel/include/arch/sparc64/types.h Sun Nov 7 19:08:41 2010 (r948) +++ trunk/openbios-devel/include/arch/sparc64/types.h Mon Nov 8 22:12:36 2010 (r949) @@ -25,6 +25,9 @@ typedef int int32_t; typedef long long int64_t; typedef long intptr_t; + +#define PRIx32 "x" +#define PRIx64 "llx" #endif /* endianess */ @@ -33,6 +36,8 @@ /* physical address */ typedef uint64_t phys_addr_t; +#define FMT_plx "%016" PRIx64 + /* cell based types */ typedef long long cell; typedef unsigned long long ucell; Modified: trunk/openbios-devel/include/arch/x86/types.h ============================================================================== --- trunk/openbios-devel/include/arch/x86/types.h Sun Nov 7 19:08:41 2010 (r948) +++ trunk/openbios-devel/include/arch/x86/types.h Mon Nov 8 22:12:36 2010 (r949) @@ -18,8 +18,11 @@ #include "autoconf.h" /* physical address: XXX theoretically 36 bits for PAE */ + typedef uint32_t phys_addr_t; +#define FMT_plx "%08" PRIx32 + /* cell based types */ typedef int32_t cell;

3 3

[commit] r948 - trunk/openbios-devel/arch/ppc/qemu
by repository service 07 Nov '10

07 Nov '10

Author: afaerber Date: Sun Nov 7 19:08:41 2010 New Revision: 948 URL: http://tracker.coreboot.org/trac/openbios/changeset/948 Log: ppc: Fix indentation Tab-indent to match surrounding code. Signed-off-by: Andreas Färber <andreas.faerber(a)web.de> Modified: trunk/openbios-devel/arch/ppc/qemu/ofmem.c Modified: trunk/openbios-devel/arch/ppc/qemu/ofmem.c ============================================================================== --- trunk/openbios-devel/arch/ppc/qemu/ofmem.c Sun Nov 7 17:51:47 2010 (r947) +++ trunk/openbios-devel/arch/ppc/qemu/ofmem.c Sun Nov 7 19:08:41 2010 (r948) @@ -395,7 +395,7 @@ /* SDR1: Storage Description Register 1 */ hash_base = (ramsize - 0x00100000 - HASH_SIZE) & hash_mask; - memset((void *)hash_base, 0, HASH_SIZE); + memset((void *)hash_base, 0, HASH_SIZE); sdr1 = hash_base | ((HASH_SIZE-1) >> 16); asm volatile("mtsdr1 %0" :: "r" (sdr1) );

1 0

[PATCH] ppc: Fix /memory reg property
by Andreas Färber 07 Nov '10

07 Nov '10

Physical address and size were both using two cells instead of one, and their order was wrong. This would happen to work for RAM < 4 GB, since the clear high address bits would happen to match the zero-based memory location. It would indicate a second bank of size zero though. Switch the value order and use encode-phys in preparation for ppc64. This tidies the output of the Haiku boot loader. Cc: Alexander Graf <agraf(a)suse.de> Signed-off-by: Andreas Färber <andreas.faerber(a)web.de> --- arch/ppc/qemu/init.c | 10 ++-------- 1 files changed, 2 insertions(+), 8 deletions(-) diff --git a/arch/ppc/qemu/init.c b/arch/ppc/qemu/init.c index bfd147a..c5da992 100644 --- a/arch/ppc/qemu/init.c +++ b/arch/ppc/qemu/init.c @@ -730,15 +730,9 @@ arch_of_init( void ) /* all memory */ - PUSH(ram_size >> 32); - fword("encode-int"); - PUSH(ram_size & 0xffffffff); - fword("encode-int"); - fword("encode+"); - PUSH(0); - fword("encode-int"); - fword("encode+"); PUSH(0); + fword("encode-phys"); + PUSH(ram_size & 0xffffffff); fword("encode-int"); fword("encode+"); push_str("reg"); -- 1.7.3

2 4

[commit] r947 - trunk/openbios-devel/arch/ppc/qemu
by repository service 07 Nov '10

07 Nov '10

Author: afaerber Date: Sun Nov 7 17:51:47 2010 New Revision: 947 URL: http://tracker.coreboot.org/trac/openbios/changeset/947 Log: ppc: Fix /memory reg property Physical address and size were both using two cells instead of one, and their order was wrong. This would happen to work for RAM < 4 GB, since the clear high address bits would happen to match the zero-based memory location. It would indicate a second bank of size zero though. Switch the value order and use encode-phys in preparation for ppc64. This tidies the output of the Haiku boot loader. v2: * Document TODOs for ppc64, requested by Alex. Cc: Alexander Graf <agraf(a)suse.de> Signed-off-by: Andreas Färber <andreas.faerber(a)web.de> Modified: trunk/openbios-devel/arch/ppc/qemu/init.c Modified: trunk/openbios-devel/arch/ppc/qemu/init.c ============================================================================== --- trunk/openbios-devel/arch/ppc/qemu/init.c Sat Nov 6 02:17:02 2010 (r946) +++ trunk/openbios-devel/arch/ppc/qemu/init.c Sun Nov 7 17:51:47 2010 (r947) @@ -730,15 +730,12 @@ /* all memory */ - PUSH(ram_size >> 32); - fword("encode-int"); - PUSH(ram_size & 0xffffffff); - fword("encode-int"); - fword("encode+"); - PUSH(0); - fword("encode-int"); - fword("encode+"); + /* TODO Adjust this when #address-cells gets increased for ppc64. */ PUSH(0); + fword("encode-phys"); + /* This needs adjusting if #size-cells gets increased. + Alternatively use multiple (address, size) tuples. */ + PUSH(ram_size & 0xffffffff); fword("encode-int"); fword("encode+"); push_str("reg");

1 0

OFMEM and physical addresses
by Mark Cave-Ayland 07 Nov '10

07 Nov '10

Hi all, I've spent a few hours looking at converting SPARC32 over to OFMEM and have hit a fairly fundamental hurdle at the start - the OFMEM code doesn't handle physical addresses properly. In terms of mapping I/O devices into address space, the first problem is with the ofmem_arch_early_map_pages() API which looks like this: ofmem_arch_early_map_pages(ucell phys, ucell virt, ucell size, ucell mode); The problem here is that when setting up the QEMU firmware configuration interface with an address of 0xd00000510, once we hit the page mapping functions we lose the top 8 bits of the physical address since sizeof(phys_addr_t) > size(ucell) on SPARC32. This causes incorrect physical address mappings. My current thinking is that all physical addresses should be changed to phys_addr_t in the OFMEM APIs, but this is slightly more complicated because the same range_t structure is used for storing both physical and virtual range mappings. I believe that if I change the OFMEM range_t structure to use phys_addr_t instead of ucell then this should work assuming that sizeof(phys_addr_t) is always >= sizeof(ucell). Looking at the OF spec it seems that all virtual addresses always fit within 1 Forth cell, and so virtual addresses should be fine in their present form. Can anyone see a flaw in the above logic before I go ahead and code it up? ATB, Mark. -- Mark Cave-Ayland - Senior Technical Architect PostgreSQL - PostGIS Sirius Corporation plc - control through freedom http://www.siriusit.co.uk t: +44 870 608 0063 Sirius Labs: http://www.siriusit.co.uk/labs

3 5