OpenBIOS March 2014

openbios@openbios.org

7 participants
30 discussions

by BALATON Zoltan

Hello, I'm sending my previous patches split into multiple patches and some more I've done since. This is not including the change of the / node name as it was suggested to preserve the original name on some other property but the name of that new property was not decided so I don't know what to do with it but that's also needed to get further with MorphOS. Regards, BALATON Zoltan

6 years, 10 months

[commit] r1283 - trunk/openbios-devel/libopenbios

by repository service

Author: mcayland Date: Sun Mar 30 19:31:19 2014 New Revision: 1283 URL: http://tracker.coreboot.org/trac/openbios/changeset/1283 Log: OFMEM: fix up debugging comment in ofmem_claim_phys(). It was mistakingly indicating ofmem_claim() as the source function rather than ofmem_claim_phys(). Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/libopenbios/ofmem_common.c Modified: trunk/openbios-devel/libopenbios/ofmem_common.c ============================================================================== --- trunk/openbios-devel/libopenbios/ofmem_common.c Sun Mar 30 19:31:03 2014 (r1282) +++ trunk/openbios-devel/libopenbios/ofmem_common.c Sun Mar 30 19:31:19 2014 (r1283) @@ -521,7 +521,7 @@ /* if align != 0, phys is ignored. Returns -1 on error */ phys_addr_t ofmem_claim_phys( phys_addr_t phys, ucell size, ucell align ) { - OFMEM_TRACE("ofmem_claim phys=" FMT_plx " size=" FMT_ucellx + OFMEM_TRACE("ofmem_claim_phys phys=" FMT_plx " size=" FMT_ucellx " align=" FMT_ucellx "\n", phys, size, align);

6 years, 11 months

[PATCH] OBP Memory deallocation

by Olivier Danet

During boot, NextSTEP do memory allocation & deallocation: - Implement dumb_memfree() - Unmap freed memory (during boot, all the virtual memory is probed, so freed memory must be properly unmapped) - The pointer given to dumb_memfree() does not always match the pointer returned by dumb_memalloc(). In that case, try to find a memory area which includes the pointer address. Ugly, indeed... Signed-off-by : Olivier Danet <odanet(a)caramail.com> Index: arch/sparc32/lib.c =================================================================== --- arch/sparc32/lib.c (révision 1257) +++ arch/sparc32/lib.c (copie de travail) @@ -239,10 +245,18 @@ return obp_memalloc(va, size, align); } -void obp_dumb_memfree(__attribute__((unused))char *va, - __attribute__((unused))unsigned sz) +void obp_dumb_memfree(char *va, unsigned size) { - DPRINTF("obp_dumb_memfree 0x%p (size %d)\n", va, sz); + phys_addr_t phys; + ucell cellmode; + + DPRINTF("obp_dumb_memfree: virta 0x%x, sz %d\n", (unsigned int)va, size); + + phys = ofmem_translate(pointer2cell(va), &cellmode); + + ofmem_unmap(pointer2cell(va), size); + ofmem_release_virt(pointer2cell(va), size); + ofmem_release_phys(phys, size); } /* Data fault handling routines */ Index: arch/sparc32/ofmem_sparc32.c =================================================================== --- arch/sparc32/ofmem_sparc32.c (révision 1257) +++ arch/sparc32/ofmem_sparc32.c (copie de travail) @@ -133,17 +133,24 @@ /* Generate memory available property entry for Sparc32 */ void ofmem_arch_create_available_entry(phandle_t ph, ucell *availentry, phys_addr_t start, ucell size) { - int i = 0; + int i; - i += ofmem_arch_encode_physaddr(availentry, start); + i = ofmem_arch_encode_physaddr(availentry, start); availentry[i] = size; } /* Unmap a set of pages */ void ofmem_arch_unmap_pages(ucell virt, ucell size) { - /* OFMEM re-maps the pages for us, so just ensure the TLB is in sync */ - srmmu_flush_whole_tlb(); + unsigned i; + unsigned long pa; + + for (i = 0; i < size; i += PAGE_SIZE) { + pa = find_pte(virt, 0); + *(uint32_t *)pa = 0; + virt += PAGE_SIZE; + } + srmmu_flush_whole_tlb(); } /* Map a set of pages */ Index: arch/sparc32/romvec.h =================================================================== --- arch/sparc32/romvec.h (révision 1257) +++ arch/sparc32/romvec.h (copie de travail) @@ -30,8 +30,8 @@ int obp_inst2pkg_handler(int dev_desc); char *obp_dumb_memalloc(char *va, unsigned int size); char *obp_dumb_memalloc_handler(char *va, unsigned int size); -void obp_dumb_memfree(__attribute__((unused))char *va, __attribute__((unused))unsigned sz); -void obp_dumb_memfree_handler(__attribute__((unused))char *va, __attribute__((unused))unsigned sz); +void obp_dumb_memfree(char *va, unsigned size); +void obp_dumb_memfree_handler(char *va, unsigned size); char *obp_dumb_mmap(char *va, int which_io, unsigned int pa, unsigned int size); char *obp_dumb_mmap_handler(char *va, int which_io, unsigned int pa, unsigned int size); void obp_dumb_munmap(__attribute__((unused)) char *va, __attribute__((unused)) unsigned int size); Index: libopenbios/ofmem_common.c =================================================================== --- libopenbios/ofmem_common.c (révision 1257) +++ libopenbios/ofmem_common.c (copie de travail) @@ -887,16 +887,25 @@ cr = *r; *r = (**r).next; free(cr); - return; } + cr = *r; + for ( ; *r && ((**r).next)->start < ea; r = &(**r).next) { + cr = *r; + } - for( ; *r && ((**r).next)->start < ea; r=&(**r).next ) { + if (((**r).next)->start == ea) { + /* Found matching entry */ + cr = (**r).next; + (**r).next = cr->next; + free(cr); + + } else { + /* Not exactly matching, delete the previous entry, with start < ea */ + free(cr->next); + cr->next = (*r)->next; + return; } - - cr = (**r).next; - (**r).next = cr->next; - free(cr); } static int remove_range( phys_addr_t ea, ucell size, range_t **r ) ===================================================================

6 years, 11 months

[commit] r1282 - trunk/openbios-devel/libopenbios

by repository service

Author: mcayland Date: Sun Mar 30 19:31:03 2014 New Revision: 1282 URL: http://tracker.coreboot.org/trac/openbios/changeset/1282 Log: OFMEM: tidy up "bad parameters" debugging message. Display the values in lower-case hex to match the rest of the debugging code. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/libopenbios/ofmem_common.c Modified: trunk/openbios-devel/libopenbios/ofmem_common.c ============================================================================== --- trunk/openbios-devel/libopenbios/ofmem_common.c Sun Mar 30 19:31:01 2014 (r1281) +++ trunk/openbios-devel/libopenbios/ofmem_common.c Sun Mar 30 19:31:03 2014 (r1282) @@ -791,7 +791,7 @@ if( (phys & (PAGE_SIZE - 1)) || (virt & (PAGE_SIZE - 1)) || (size & (PAGE_SIZE - 1)) ) { OFMEM_TRACE("ofmem_map: Bad parameters (" - FMT_plx " " FMT_ucellX " " FMT_ucellX ")\n", + FMT_plx " " FMT_ucellx " " FMT_ucellx ")\n", phys, virt, size ); phys &= PAGE_MASK;

6 years, 11 months

[commit] r1281 - trunk/openbios-devel/libopenbios

by repository service

Author: mcayland Date: Sun Mar 30 19:31:01 2014 New Revision: 1281 URL: http://tracker.coreboot.org/trac/openbios/changeset/1281 Log: OFMEM: improve remove_range_() function algorithm Rework the algorithm used in remove_range_() in order to make it similar to the algorithm used in unmap_pages(). This has several benefits: in particular the unmap_pages()/remove_range_() functions now have the same behaviour, the code is noticeably simpler, and it also correctly handles the case where the start address of the range to be removed doesn't match the range exactly (such as occurs when attempting to boot NextStep under SPARC32). Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/libopenbios/ofmem_common.c Modified: trunk/openbios-devel/libopenbios/ofmem_common.c ============================================================================== --- trunk/openbios-devel/libopenbios/ofmem_common.c Sun Mar 9 21:51:13 2014 (r1280) +++ trunk/openbios-devel/libopenbios/ofmem_common.c Sun Mar 30 19:31:01 2014 (r1281) @@ -885,23 +885,18 @@ static void remove_range_( phys_addr_t ea, ucell size, range_t **r ) { - range_t *cr; + range_t **t, *u; - /* Handle special case if we're removing the first entry */ - if ((**r).start >= ea) { - cr = *r; - *r = (**r).next; - free(cr); - - return; - } - - for( ; *r && ((**r).next)->start < ea; r=&(**r).next ) { + for (t = r; *t; t = &(**t).next) { + if (ea >= (**t).start && ea + size <= (**t).start + (**t).size) { + OFMEM_TRACE("remove_range_ freeing range with addr=" FMT_plx + " size=" FMT_ucellx "\n", (**t).start, (**t).size); + u = *t; + *t = (**t).next; + free(u); + break; + } } - - cr = (**r).next; - (**r).next = cr->next; - free(cr); } static int remove_range( phys_addr_t ea, ucell size, range_t **r )

6 years, 11 months

[commit] r1280 - in trunk/openbios-devel: forth/debugging libopenbios packages

by repository service

Author: mcayland Date: Sun Mar 9 21:51:13 2014 New Revision: 1280 URL: http://tracker.coreboot.org/trac/openbios/changeset/1280 Log: bootcode_load.c: use new get-bootcode-info plus minor bootcode fixes Bring the bootcode loader in line with all of the other existing loaders, making use of our new get-bootcode-info word to pass the parameters instead of creating global variables for them all. Here we remove all global variables and references to the fixed load address used by the quik bootcode loader as this is handled by get-bootcode-info. This exposes a minor bug where we need to return success as soon as a valid bootcode is detected; otherwise we inadvertently drop into the filesystem detection code which is invalid. Based upon a patch by Andrei E. Warkentin <andrey.warkentin(a)gmail.com>. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/forth/debugging/client.fs trunk/openbios-devel/libopenbios/bootcode_load.c trunk/openbios-devel/libopenbios/load.c trunk/openbios-devel/packages/mac-parts.c Modified: trunk/openbios-devel/forth/debugging/client.fs ============================================================================== --- trunk/openbios-devel/forth/debugging/client.fs Sun Mar 9 21:50:51 2014 (r1279) +++ trunk/openbios-devel/forth/debugging/client.fs Sun Mar 9 21:51:13 2014 (r1280) @@ -27,9 +27,6 @@ variable state-valid 0 state-valid ! -variable want-bootcode -0 want-bootcode ! - variable file-size : !load-size file-size ! ; Modified: trunk/openbios-devel/libopenbios/bootcode_load.c ============================================================================== --- trunk/openbios-devel/libopenbios/bootcode_load.c Sun Mar 9 21:50:51 2014 (r1279) +++ trunk/openbios-devel/libopenbios/bootcode_load.c Sun Mar 9 21:51:13 2014 (r1280) @@ -12,13 +12,13 @@ #define printf printk #define debug printk -#define OLDWORLD_BOOTCODE_BASEADDR (0x3f4000) int bootcode_load(ihandle_t dev) { int retval = -1, count = 0, fd; - unsigned long bootcode, loadbase, offset; + unsigned long bootcode, loadbase, entry, size, offset; + ihandle_t bootcode_info; /* Mark the saved-program-state as invalid */ feval("0 state-valid !"); @@ -28,18 +28,37 @@ goto out; } + /* If we don't have the get-bootcode-info word then we don't support + loading bootcode via %BOOT */ + bootcode_info = find_ih_method("get-bootcode-info", dev); + if (!bootcode_info) { + goto out; + } + /* Default to loading at load-base */ fword("load-base"); loadbase = POP(); + entry = loadbase; + size = 0; #ifdef CONFIG_PPC - /* ...except that QUIK (the only known user of %BOOT to date) is built - with a hard-coded address of 0x3f4000. Let's just use this for the - moment on both New World and Old World Macs, allowing QUIK to also - work under a New World Mac. If we find another user of %BOOT we can - rethink this later. PReP machines should be left unaffected. */ + /* + * Apple OF does not honor load-base and instead uses pmBootLoad + * value from the boot partition descriptor. + * + * Tested with: + * a debian image with QUIK installed + * a debian image with iQUIK installed (https://github.com/andreiw/quik) + * an IQUIK boot floppy + * a NetBSD boot floppy (boots stage 2) + */ if (is_apple()) { - loadbase = OLDWORLD_BOOTCODE_BASEADDR; + PUSH(bootcode_info); + fword("execute"); + + loadbase = POP(); + entry = POP(); + size = POP(); } #endif @@ -58,11 +77,16 @@ if (!count) { goto out; } + + /* Use proper file size if we got it from bootcode info */ + if (size == 0) { + size = offset; + } /* Initialise saved-program-state */ - PUSH(loadbase); + PUSH(entry); feval("saved-program-state >sps.entry !"); - PUSH(offset); + PUSH(size); feval("saved-program-state >sps.file-size !"); feval("bootcode saved-program-state >sps.file-type !"); Modified: trunk/openbios-devel/libopenbios/load.c ============================================================================== --- trunk/openbios-devel/libopenbios/load.c Sun Mar 9 21:50:51 2014 (r1279) +++ trunk/openbios-devel/libopenbios/load.c Sun Mar 9 21:51:13 2014 (r1280) @@ -108,11 +108,13 @@ #ifdef CONFIG_LOADER_BOOTCODE /* Check for a "raw" %BOOT bootcode payload */ - feval("want-bootcode @"); - valid = POP(); - if (valid) { - bootcode_load(dev); - } + bootcode_load(dev); + feval("state-valid @"); + valid = POP(); + if (valid) { + feval("saved-program-state >sps.file-size @"); + return; + } #endif /* Didn't load anything, so return zero size */ Modified: trunk/openbios-devel/packages/mac-parts.c ============================================================================== --- trunk/openbios-devel/packages/mac-parts.c Sun Mar 9 21:50:51 2014 (r1279) +++ trunk/openbios-devel/packages/mac-parts.c Sun Mar 9 21:51:13 2014 (r1280) @@ -92,9 +92,6 @@ /* Detect if we are looking for the bootcode */ if (strcmp(argstr, "%BOOT") == 0) { want_bootcode = 1; - feval("1 want-bootcode !"); - } else { - feval("0 want-bootcode !"); } } @@ -261,6 +258,11 @@ di->size_hi = size >> BITS; di->size_lo = size & (ucell) -1; + /* If we're trying to execute bootcode then we're all done */ + if (want_bootcode) { + goto out; + } + /* We have a valid partition - so probe for a filesystem at the current offset */ DPRINTF("mac-parts: about to probe for fs\n"); DPUSH( offs ); @@ -289,7 +291,7 @@ /* If we have been asked to open a particular file, interpose the filesystem package with the passed filename as an argument */ - if (!want_bootcode && strlen(argstr)) { + if (strlen(argstr)) { push_str( argstr ); PUSH_ph( ph ); fword("interpose");

6 years, 11 months

[commit] r1279 - trunk/openbios-devel/packages

by repository service

Author: mcayland Date: Sun Mar 9 21:50:51 2014 New Revision: 1279 URL: http://tracker.coreboot.org/trac/openbios/changeset/1279 Log: mac-parts.c: introduce new get-bootcode-info word This is an intermediate step in preparation for allowing %BOOT to read its load/entry addresses from the partition header, based upon an original patch by Andrei E. Warkentin <andrey.warkentin(a)gmail.com>. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/packages/mac-parts.c Modified: trunk/openbios-devel/packages/mac-parts.c ============================================================================== --- trunk/openbios-devel/packages/mac-parts.c Sun Mar 9 21:50:36 2014 (r1278) +++ trunk/openbios-devel/packages/mac-parts.c Sun Mar 9 21:50:51 2014 (r1279) @@ -35,6 +35,7 @@ xt_t seek_xt, read_xt; ucell offs_hi, offs_lo; ucell size_hi, size_lo; + ucell bootcode_addr, bootcode_entry; unsigned int blocksize; phandle_t filesystem_ph; } macparts_info_t; @@ -237,8 +238,19 @@ size = (long long)__be32_to_cpu(par.pmPartBlkCnt) * bs; if (want_bootcode) { - offs += (long long)__be32_to_cpu(par.pmLgBootStart) * bs; + + /* If size == 0 then fail because we requested bootcode but it doesn't exist */ size = (long long)__be32_to_cpu(par.pmBootSize); + if (!size) { + ret = 0; + goto out; + } + + /* Adjust seek position so 0 = start of bootcode */ + offs += (long long)__be32_to_cpu(par.pmLgBootStart) * bs; + + di->bootcode_addr = __be32_to_cpu(par.pmBootLoad); + di->bootcode_entry = __be32_to_cpu(par.pmBootEntry); } di->blocksize = (unsigned int)bs; @@ -326,6 +338,17 @@ PUSH( di->size_hi ); } +/* ( -- size entry addr ) */ +static void +macparts_get_bootcode_info( macparts_info_t *di ) +{ + DPRINTF("macparts_get_bootcode_info"); + + PUSH( di->size_lo ); + PUSH( di->bootcode_entry ); + PUSH( di->bootcode_addr ); +} + static void macparts_block_size( macparts_info_t *di ) { @@ -403,15 +426,16 @@ } NODE_METHODS( macparts ) = { - { "probe", macparts_probe }, - { "open", macparts_open }, - { "seek", macparts_seek }, - { "read", macparts_read }, - { "load", macparts_load }, - { "dir", macparts_dir }, - { "get-info", macparts_get_info }, - { "block-size", macparts_block_size }, - { NULL, macparts_initialize }, + { "probe", macparts_probe }, + { "open", macparts_open }, + { "seek", macparts_seek }, + { "read", macparts_read }, + { "load", macparts_load }, + { "dir", macparts_dir }, + { "get-info", macparts_get_info }, + { "get-bootcode-info", macparts_get_bootcode_info }, + { "block-size", macparts_block_size }, + { NULL, macparts_initialize }, }; void

6 years, 11 months

[commit] r1278 - trunk/openbios-devel/packages

by repository service

Author: mcayland Date: Sun Mar 9 21:50:36 2014 New Revision: 1278 URL: http://tracker.coreboot.org/trac/openbios/changeset/1278 Log: mac-parts.c: prevent hang when opening partition without valid filesystem If a partition was opened with a file argument such as "hd:2,\\:tbxi" and no valid filesystem was detected, mac-parts would still attempt to access the file system in order to evaluate the file argument resulting in a hang. Make sure that we return a failure code to the caller to ensure that OpenBIOS doesn't continue to try and access the file when the underlying filesystem is invalid. As reported by Andrei E. Warkentin <andrey.warkentin(a)gmail.com>. Signed-off-by: Mark Cave-Ayland <mark.cave-ayland(a)ilande.co.uk> Modified: trunk/openbios-devel/packages/mac-parts.c Modified: trunk/openbios-devel/packages/mac-parts.c ============================================================================== --- trunk/openbios-devel/packages/mac-parts.c Sat Mar 8 14:03:01 2014 (r1277) +++ trunk/openbios-devel/packages/mac-parts.c Sun Mar 9 21:50:36 2014 (r1278) @@ -286,6 +286,12 @@ goto out; } else { DPRINTF("mac-parts: no filesystem found on partition %d; bypassing misc-files interpose\n", parnum); + + /* Here we have a valid partition; however if we tried to pass in a file argument for a + partition that doesn't contain a filesystem, then we must fail */ + if (strlen(argstr)) { + ret = 0; + } } }

6 years, 11 months

Re: [OpenBIOS] [PATCH 0/4] Changes for MorphOS

by BALATON Zoltan

On Sat Mar 8 2014, Mark Cave-Ayland wrote: > I think the best solution is going to be to move the default device name > from forth/device/tree.fs into the individual per-architecture > arch/*/tree.fs files and then customise the PPC one as appropriate. Is it possible to set/override the value in forth/device/tree.fs in the ppc/qemu/init.c file somehow? If so then it would be easier and we already have an is_apple() function there so this can be done only for Apple emulation. Is the init code run when the tree is already set up? What should it do to replace the name property? A assume it would call some forth code to do this but I don't know how to do this properly. Any ideas? Regards, BALATON Zoltan

6 years, 11 months

Re: [OpenBIOS] [PATCH 0/4] Changes for MorphOS

by BALATON Zoltan

> I missed the reason you wanted the name changed; I commented at the time > that for IEEE 1275, the name property at the root node is supposed to be > the platform name. Why do you want it called "device-tree", which would > certainly be incorrect? To match Apple's implementation which is expected my at least one client. Sould I enclose it in CONFIG_PPC or some other ifdefs? Would that make it better? Regards, BALATON Zoltan

6 years, 11 months