coreboot August 2008

coreboot@coreboot.org

64 participants
593 discussions

[PATCH] v3: prepare stage2 for -fwhole-program
by Carl-Daniel Hailfinger 28 Aug '08

28 Aug '08

gcc -fwhole-program needs to mark any function which is used outside the current compilation unit to be marked as externally_visible. We have EXPORT_SYMBOL exactly for that purpose. This applies to the following symbols used by x86emu and/or vm86: - pci_read_config8 - pci_read_config16 - pci_read_config32 - pci_write_config8 - pci_write_config16 - pci_write_config32 - dev_find_pci_device - dev_find_slot It also applies to the main entry point of stage2: - stage2 With this patch, I can use -fwohle-program for stage2 without any problems. For standard compilation, this is a noop. Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006(a)gmx.net> Index: corebootv3-makefilerewrite-fwholeprogram/include/device/pci_ops.h =================================================================== --- corebootv3-makefilerewrite-fwholeprogram/include/device/pci_ops.h (Revision 824) +++ corebootv3-makefilerewrite-fwholeprogram/include/device/pci_ops.h (Arbeitskopie) @@ -24,11 +24,17 @@ #include <shared.h> u8 pci_read_config8(struct device * dev, unsigned where); +EXPORT_SYMBOL(pci_read_config8); u16 pci_read_config16(struct device * dev, unsigned where); +EXPORT_SYMBOL(pci_read_config16); u32 pci_read_config32(struct device * dev, unsigned where); +EXPORT_SYMBOL(pci_read_config32); void pci_write_config8(struct device * dev, unsigned where, u8 val); +EXPORT_SYMBOL(pci_write_config8); void pci_write_config16(struct device * dev, unsigned where, u16 val); +EXPORT_SYMBOL(pci_write_config16); void pci_write_config32(struct device * dev, unsigned where, u32 val); +EXPORT_SYMBOL(pci_write_config32); u8 pci_conf1_read_config8(u32 bdf, int where); EXPORT_SYMBOL(pci_conf1_read_config8); Index: corebootv3-makefilerewrite-fwholeprogram/include/device/device.h =================================================================== --- corebootv3-makefilerewrite-fwholeprogram/include/device/device.h (Revision 824) +++ corebootv3-makefilerewrite-fwholeprogram/include/device/device.h (Arbeitskopie) @@ -24,6 +24,7 @@ #include <types.h> #include <device/resource.h> #include <device/path.h> +#include <shared.h> /** * Create a 32-bit value from four characters. This is better @@ -268,8 +269,10 @@ struct device * alloc_find_dev(struct bus *parent, struct device_path *path, struct device_id *id); struct device * dev_find_device (struct device_id *devid, struct device * from); struct device *dev_find_pci_device(u16 vendor, u16 device, struct device *from); +EXPORT_SYMBOL(dev_find_pci_device); struct device * dev_find_class (unsigned int class, struct device * from); struct device * dev_find_slot (unsigned int bus, unsigned int devfn); +EXPORT_SYMBOL(dev_find_slot); struct device * dev_find_slot_on_smbus (unsigned int bus, unsigned int addr); void default_device_constructor(struct device *dev, struct device_operations *constructor); Index: corebootv3-makefilerewrite-fwholeprogram/lib/stage2.c =================================================================== --- corebootv3-makefilerewrite-fwholeprogram/lib/stage2.c (Revision 824) +++ corebootv3-makefilerewrite-fwholeprogram/lib/stage2.c (Arbeitskopie) @@ -93,3 +93,4 @@ return 0; } +EXPORT_SYMBOL(stage2); -- http://www.hailfinger.org/

2 2

r827 - in coreboot-v3: include/device lib
by svn＠coreboot.org 28 Aug '08

28 Aug '08

Author: hailfinger Date: 2008-08-28 00:29:38 +0200 (Thu, 28 Aug 2008) New Revision: 827 Modified: coreboot-v3/include/device/device.h coreboot-v3/include/device/pci_ops.h coreboot-v3/lib/stage2.c Log: gcc -fwhole-program needs to mark any function which is used outside the current compilation unit to be marked as externally_visible. We have EXPORT_SYMBOL exactly for that purpose. This applies to the following symbols used by x86emu and/or vm86: - pci_read_config8 - pci_read_config16 - pci_read_config32 - pci_write_config8 - pci_write_config16 - pci_write_config32 - dev_find_pci_device - dev_find_slot It also applies to the main entry point of stage2: - stage2 With this patch, I can use -fwhole-program for stage2 without any problems. For standard compilation, this is a noop. Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006(a)gmx.net> Acked-by: Ronald G. Minnich <rminnich(a)gmail.com> Modified: coreboot-v3/include/device/device.h =================================================================== --- coreboot-v3/include/device/device.h 2008-08-27 05:30:50 UTC (rev 826) +++ coreboot-v3/include/device/device.h 2008-08-27 22:29:38 UTC (rev 827) @@ -24,6 +24,7 @@ #include <types.h> #include <device/resource.h> #include <device/path.h> +#include <shared.h> /** * Create a 32-bit value from four characters. This is better @@ -268,8 +269,10 @@ struct device * alloc_find_dev(struct bus *parent, struct device_path *path, struct device_id *id); struct device * dev_find_device (struct device_id *devid, struct device * from); struct device *dev_find_pci_device(u16 vendor, u16 device, struct device *from); +EXPORT_SYMBOL(dev_find_pci_device); struct device * dev_find_class (unsigned int class, struct device * from); struct device * dev_find_slot (unsigned int bus, unsigned int devfn); +EXPORT_SYMBOL(dev_find_slot); struct device * dev_find_slot_on_smbus (unsigned int bus, unsigned int addr); void default_device_constructor(struct device *dev, struct device_operations *constructor); Modified: coreboot-v3/include/device/pci_ops.h =================================================================== --- coreboot-v3/include/device/pci_ops.h 2008-08-27 05:30:50 UTC (rev 826) +++ coreboot-v3/include/device/pci_ops.h 2008-08-27 22:29:38 UTC (rev 827) @@ -24,11 +24,17 @@ #include <shared.h> u8 pci_read_config8(struct device * dev, unsigned where); +EXPORT_SYMBOL(pci_read_config8); u16 pci_read_config16(struct device * dev, unsigned where); +EXPORT_SYMBOL(pci_read_config16); u32 pci_read_config32(struct device * dev, unsigned where); +EXPORT_SYMBOL(pci_read_config32); void pci_write_config8(struct device * dev, unsigned where, u8 val); +EXPORT_SYMBOL(pci_write_config8); void pci_write_config16(struct device * dev, unsigned where, u16 val); +EXPORT_SYMBOL(pci_write_config16); void pci_write_config32(struct device * dev, unsigned where, u32 val); +EXPORT_SYMBOL(pci_write_config32); u8 pci_conf1_read_config8(u32 bdf, int where); EXPORT_SYMBOL(pci_conf1_read_config8); Modified: coreboot-v3/lib/stage2.c =================================================================== --- coreboot-v3/lib/stage2.c 2008-08-27 05:30:50 UTC (rev 826) +++ coreboot-v3/lib/stage2.c 2008-08-27 22:29:38 UTC (rev 827) @@ -93,3 +93,4 @@ return 0; } +EXPORT_SYMBOL(stage2);

1 0

[PATCH] Inefficient test in buildtarget script
by Mats Erik Andersson 28 Aug '08

28 Aug '08

Capture output from 'ld --help', in order that the test becomes meaningful. The linker puts help texts on stderr. Signed-off-by: Mats Erik Andersson <mats.andersson(a)gisladisker.se> --- Index: targets/buildtarget =================================================================== --- targets/buildtarget (revision 3531) +++ targets/buildtarget (arbetskopia) @@ -69,7 +69,7 @@ rm -rf .$$.tmp -if $CC -Wl,--help | grep -q build-id; then +if $CC -Wl,--help 2>&1 | grep -q build-id; then EXTRA_LFLAGS="$EXTRA_LFLAGS -Wl,--build-id=none" fi

3 4

r3543 - trunk/coreboot-v2/targets
by svn＠coreboot.org 28 Aug '08

28 Aug '08

Author: hailfinger Date: 2008-08-28 00:25:55 +0200 (Thu, 28 Aug 2008) New Revision: 3543 Modified: trunk/coreboot-v2/targets/buildtarget Log: Capture output from 'ld --help', in order that the test becomes meaningful. The linker puts help texts on stderr. Signed-off-by: Mats Erik Andersson <mats.andersson(a)gisladisker.se> Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006(a)gmx.net> Modified: trunk/coreboot-v2/targets/buildtarget =================================================================== --- trunk/coreboot-v2/targets/buildtarget 2008-08-27 21:53:11 UTC (rev 3542) +++ trunk/coreboot-v2/targets/buildtarget 2008-08-27 22:25:55 UTC (rev 3543) @@ -69,7 +69,12 @@ rm -rf .$$.tmp -if $CC -Wl,--help | grep -q build-id; then +# The linker output fd depends on the gcc version. +# 1) 'ld-2.15 --help' and 'gcc-4.0 -Wl,--help' use STDOUT. +# 2) 'gcc-3.3 --help' and 'gcc-3.4 -Wl,--help' use STDERR. +# Thus older versions of GCC (presumably 3.x) implement a redirection of +# output to stderr. Re-redirect stderr to stdout to work always. +if $CC -Wl,--help 2>&1 | grep -q build-id; then EXTRA_LFLAGS="$EXTRA_LFLAGS -Wl,--build-id=none" fi

1 0

r3542 - in trunk/coreboot-v2/src: lib stream
by svn＠coreboot.org 27 Aug '08

27 Aug '08

Author: ward Date: 2008-08-27 23:53:11 +0200 (Wed, 27 Aug 2008) New Revision: 3542 Modified: trunk/coreboot-v2/src/lib/lzma.c trunk/coreboot-v2/src/stream/rom_stream.c Log: If you have option CONFIG_COMPRESSED_PAYLOAD_LZMA=1 option CONFIG_PRECOMPRESSED_PAYLOAD=1 set in Config.lb but accidentally use an uncompressed payload, coreboot (v2) bombs out like this: elfboot: Attempting to load payload. rom_stream: 0xfffc0000 - 0xfffdefff Uncompressing to RAM 0x01000000 Decoder scratchpad too small! Decoding error = 1 Unexpected Exception: 6 @ 10:04000408 - Halting Code: 0 eflags: 00010057 eax: 00000101 ebx: 04000400 ecx: 000003d4 edx: fffc0000 edi: 04000400 esi: 04000401 ebp: 04000400 esp: 0013dfb4 The attached patch modifies v2's lzma code so that it assumes an uncompressed payload if it fails to find a properly compressed payload. Compare with the fatal error above: elfboot: Attempting to load payload. rom_stream: 0xfffc0000 - 0xfffdefff Uncompressing to RAM 0x01000000 Decoder scratchpad too small! olen = 0x00000000 done. Decompression failed. Assuming payload is uncompressed... Found ELF candidate at offset 0 header_offset is 0 Try to load at offset 0x0 If you don't have CONFIG_COMPRESSED_PAYLOAD_LZMA and CONFIG_PRECOMPRESSED_PAYLOAD set and use an uncompressed payload, things are as before: elfboot: Attempting to load payload. rom_stream: 0xfffc0000 - 0xfffdefff Found ELF candidate at offset 0 header_offset is 0 Try to load at offset 0x0 One can argue that this is a case of 'builder beware', but my counter argument is that anything that causes unexpected runtime breakage is really, really, really bad, and should be avoided where possible. This patch also fixes one erroneous comment. Signed-off-by: Ward Vandewege <ward(a)gnu.org> Acked-by: Myles Watson <mylesgw(a)gmail.com> Acked-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006(a)gmx.net> Modified: trunk/coreboot-v2/src/lib/lzma.c =================================================================== --- trunk/coreboot-v2/src/lib/lzma.c 2008-08-27 21:28:41 UTC (rev 3541) +++ trunk/coreboot-v2/src/lib/lzma.c 2008-08-27 21:53:11 UTC (rev 3542) @@ -28,16 +28,19 @@ outSize = *(UInt32 *)(src + LZMA_PROPERTIES_SIZE); if (LzmaDecodeProperties(&state.Properties, properties, LZMA_PROPERTIES_SIZE) != LZMA_RESULT_OK) { printk_warning("Incorrect stream properties\n"); + return 0; } mallocneeds = (LzmaGetNumProbs(&state.Properties) * sizeof(CProb)); if (mallocneeds > 15980) { printk_warning("Decoder scratchpad too small!\n"); + return 0; } state.Probs = (CProb *)scratchpad; res = LzmaDecode(&state, src + LZMA_PROPERTIES_SIZE + 8, (SizeT)0xffffffff, &inProcessed, dst, outSize, &outProcessed); if (res != 0) { printk_warning("Decoding error = %d\n", res); + return 0; } return outSize; } Modified: trunk/coreboot-v2/src/stream/rom_stream.c =================================================================== --- trunk/coreboot-v2/src/stream/rom_stream.c 2008-08-27 21:28:41 UTC (rev 3541) +++ trunk/coreboot-v2/src/stream/rom_stream.c 2008-08-27 21:53:11 UTC (rev 3542) @@ -26,7 +26,7 @@ #error "You're defining more than one compression type, which is not allowed (of course)" #endif #define HAVE_UNCOMPRESSER 1 -// include generic nrv2b +// include generic lzma #include "../lib/lzma.c" #endif @@ -97,8 +97,14 @@ printk_debug("Uncompressing to RAM 0x%08lx ", dest); olen = uncompress((uint8_t *) rom_start, (uint8_t *)dest ); printk_debug(" olen = 0x%08lx done.\n", olen); - rom_end = dest + olen - 1; - rom = dest; + if (olen != 0) { + rom_end = dest + olen - 1; + rom = dest; + } else { + /* Decompression failed, assume payload is uncompressed */ + printk_debug("Decompression failed. Assuming payload is uncompressed...\n"); + rom = rom_start; + } #else rom = rom_start; #endif

1 0

r3541 - trunk/util/flashrom
by svn＠coreboot.org 27 Aug '08

27 Aug '08

Author: stuge Date: 2008-08-27 23:28:41 +0200 (Wed, 27 Aug 2008) New Revision: 3541 Modified: trunk/util/flashrom/flashchips.c Log: flashrom: SST25VF016B TEST_OK_ PROBE READ ERASE WRITE Per test report from Ward. Signed-off-by: Peter Stuge <peter(a)stuge.se> Acked-by: Peter Stuge <peter(a)stuge.se> Modified: trunk/util/flashrom/flashchips.c =================================================================== --- trunk/util/flashrom/flashchips.c 2008-08-27 12:53:47 UTC (rev 3540) +++ trunk/util/flashrom/flashchips.c 2008-08-27 21:28:41 UTC (rev 3541) @@ -67,7 +67,7 @@ {"PMC", "Pm49FL004", PMC_ID_NOPREFIX,PMC_49FL004, 512, 64 * 1024, TEST_OK_PREW, probe_49fl00x, erase_49fl00x, write_49fl00x}, {"Sharp", "LHF00L04", SHARP_ID, SHARP_LHF00L04, 1024, 64 * 1024, TEST_UNTESTED, probe_lhf00l04, erase_lhf00l04, write_lhf00l04}, {"Spansion", "S25FL016A", SPANSION_ID, SPANSION_S25FL016A, 2048, 256, TEST_OK_PREW, probe_spi_rdid, spi_chip_erase_c7, spi_chip_write, spi_chip_read}, - {"SST", "SST25VF016B", SST_ID, SST_25VF016B, 2048, 256, TEST_UNTESTED, probe_spi_rdid, spi_chip_erase_c7, spi_chip_write, spi_chip_read}, + {"SST", "SST25VF016B", SST_ID, SST_25VF016B, 2048, 256, TEST_OK_PREW, probe_spi_rdid, spi_chip_erase_c7, spi_chip_write, spi_chip_read}, {"SST", "SST25VF040B", SST_ID, SST_25VF040B, 512, 256, TEST_UNTESTED, probe_spi_rdid, spi_chip_erase_c7, spi_chip_write, spi_chip_read}, {"SST", "SST28SF040A", SST_ID, SST_28SF040, 512, 256, TEST_UNTESTED, probe_28sf040, erase_28sf040, write_28sf040}, {"SST", "SST29EE010", SST_ID, SST_29EE010, 128, 128, TEST_OK_PREW, probe_jedec, erase_chip_jedec, write_jedec},

1 0

romcc dislikes dump_spd_registers!
by Mats Erik Andersson 27 Aug '08

27 Aug '08

Hello again, I just realized that today's compilation efforts for my mainboard fail exactly when I include dump_spd_registers in the code. Is this known somehow? I really want to know the content in the SPD register, since I want to have a go at probing the amount of ram present on the board. When my code include a call to dump_spd_registers, the build process comes to halt due to a shortage of processor registers for the code that romcc produces. This shows up in precisely the same manner as the error that I asked about earlier today within the test suite for romcc. Best regards Mats Erik Andersson By the way, I work on a port to MS-6147/i440BX/i81371EB/W83977TF.

2 1

My romcc needs verification.
by Mats Erik Andersson 27 Aug '08

27 Aug '08

Hello all, I could use some pointers on romcc. About ten to twelve rom-images have emanated on my workstation and the target boots as it should, but a few times (not every) the output from the emulator has contained pointers on invalid code, such as: 0002:2c5f: 00 ILLEGAL EXTENDED X86 OPCODE! or 0000:4002: 01 ILLEGAL EXTENDED X86 OPCODE! Now I took to the measure of testing romcc with the delivered test suite. Most unfortunately this fails as the trailing text demonstrates. My question is this: Is it me that cannot figure things out properly regarding the use of romcc? Is it gcc-3.3 that is falling short, or does Coreboot apply tailored options to romcc, options that I did not find so far. Best regards, Mats Erik Andersson --- /* Simplified output from test phase utility for romcc. */ as tests/simple_test25.S-O2-mmmx -o tests/simple_test25.o ld -T tests/ldscript.ld tests/simple_test25.o -o tests/simple_test25.elf export ALLOC_CHECK_=2; ./romcc -fmax-allocation-passes=8 -fdebug-live-range-conflicts \ -Itests/include -O2 -mmmx -o tests/simple_test26.S-O2-mmmx \ tests/simple_test26.c > tests/simple_test26.debug pass: 0 split_ranges 0 copy 0x80ae080 canidate: 0x80af250 copy regcm: 77c 178 .... pass: 8 split_ranges 8 copy 0x80ae080 simple_test26.c:5.32: warning: edge reg %mm7 .... <built-in>:1.0: simple_test26.c:20.26: too few registers make: *** [tests/simple_test26.S-O2-mmmx] Fel 1

2 1

[PATCH] libpayload: Add Doxygen support + config file
by Uwe Hermann 27 Aug '08

27 Aug '08

See patch. Uwe. -- http://www.hermann-uwe.de | http://www.holsham-traders.de http://www.crazy-hacks.org | http://www.unmaintained-free-software.org

3 4

Re: [coreboot] [PATCH] Inefficient test in buildtarget script
by Mats Erik Andersson 27 Aug '08

27 Aug '08

Hello again, on the workstation were I did the development, the used gcc is 3.3.5 and ld is 2.15. The remarkable thing is that 1) 'ld-2.15 --help' and 'gcc-4.0 -Wl,--help' use STDOUT. 2) 'gcc-3.3 --help' and 'gcc-3.4 -Wl,--help' use STDERR. Thus older versions of GCC (presumably 3.*) implement a redirection of output. Regards Mats E A > Hi Mats, > > I would commit this right away, but on my machine, the gcc/linker > combination doesn't output anything on stderr. Could you tell us the > output of > gcc -Wl,--version > gcc --version > > I'll add that info to the script and commit. > > On 27.08.2008 14:23, Mats Erik Andersson wrote: > > Capture output from 'ld --help', in order that the test > > becomes meaningful. The linker puts help texts on stderr. > > > > Signed-off-by: Mats Erik Andersson <mats.andersson(a)gisladisker.se> > > > > Regards, > Carl-Daniel >

1 0

← Newer
1
...
9
10
11
12
13
14
15
...
60
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

coreboot August 2008