SeaBIOS

seabios@seabios.org

5 participants
3802 discussions

SeaVGABIOS with native vga init (Need testers)
by Kevin O'Connor 19 Feb '14

19 Feb '14

I have recently put together a basic implementation of a vgabios for use with displays that are initialized by coreboot. The idea is pretty simple - coreboot initializes the vga hardware and maps a framebuffer into an address in system memory. Then SeaVGABIOS implements the various legacy vga bios functions, but all it does with them is translate them to framebuffer updates. SeaVGABIOS doesn't try to initialize the display, change modes, or modify the vga registers. This might be useful for those wishing to run fewer third-party blobs, for those where a vgabios isn't available at all, and for those wanting a faster boot (I've found proprietary vga bioses to be very slow). I've run some basic tests on this version of SeaVGABIOS on my hardware. However, I don't have one of the boards where coreboot supports native vga init. It would be great if someone with the hardware (looks like stout, x60, and maybe others) could run tests. To build this version of SeaVGABIOS, checkout the testing code from: https://github.com/KevinOConnor/seabios/tree/testing Then run "make menuconfig", turn on building for coreboot (CONFIG_COREBOOT), serial debugging (CONFIG_DEBUG_SERIAL), and enable the coreboot vgabios build (CONFIG_VGA_COREBOOT). The resulting vgabios will be in "out/vgabios.bin" - this file should be placed in CBFS in the file "vgaroms/vgabios.bin". Then launch coreboot with seabios as the payload (one can use the repo above to build seabios also, or use the normal coreboot method of building seabios). Ideally, the seabios menu should appear on the screen and basic OS boot loaders will work. But if not, the debugging info should help track down any problems. -Kevin

5 8

Re: [SeaBIOS] [coreboot] SeaVGABIOS with native vga init (Need testers)
by Kevin O'Connor 19 Feb '14

19 Feb '14

On Wed, Feb 12, 2014 at 07:23:14PM -0500, Kevin O'Connor wrote: > On Wed, Feb 12, 2014 at 10:35:08PM +0100, Paul Menzel wrote: > > Am Mittwoch, den 12.02.2014, 12:58 -0500 schrieb Kevin O'Connor: > > > I've run some basic tests on this version of SeaVGABIOS on my > > > hardware. However, I don't have one of the boards where coreboot > > > supports native vga init. It would be great if someone with the > > > hardware (looks like stout, x60, and maybe others) could run tests. > > > > Thanks to Luc, the VIA K8x890 based boards like Asus M2V-MX SE have had > > native graphics init for a long time. > > > > commit aeb6c9870f0b1af8c0b55b2034f881da6757c4a4 > > Author: Luc Verhaegen <libv(a)skynet.be> > > Date: Thu Jul 23 16:04:58 2009 +0000 > > > > sb/via/k8t890: add vga textmode code for k8m890 chrome igp. > > > > Add initialisation for the VIA Chrome 9 IGP on the k8m890 through native code > > and through the general vga infrastructure i committed a month or two ago. > > Add videoram_size option for k8m890 and the Asus M2V-MX SE. > > > > Looking through the output of > > > > $ git grep k8t890 src/mainboard/ > > > > the boards Asus A8V-E Deluxe, Asus A8V-E SE, Asus K8V-X, Asus M2V-MX SE > > and Asus M2V should theoretically support that. > > Okay. The SeaVGABIOS implementation is expecting a coreboot table > (LB_TAG_FRAMEBUFFER). So, as long as that is present it should work. FYI, I updated the test vgabios code to also support "EGA style" text buffers. If the LB_TAG_FRAMEBUFFER table isn't found, then SeaVGABIOS will assume there is an EGA style text buffer. If anyone wants to give one of the above boards a try, the updated code is at: https://github.com/KevinOConnor/seabios/tree/testing Please see my previous email in this thread for directions on how to compile and use this branch. -Kevin Subject: [PATCH] vgabios: Add support for native coreboot EGA style text consoles. To: seabios(a)seabios.org If a CB_TAG_FRAMEBUFFER coreboot table isn't found, assume there is a standard text mode console mapped at 0xB8000. Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- vgasrc/Kconfig | 3 ++- vgasrc/cbvga.c | 50 ++++++++++++++++++++++++++++++++++---------------- 2 files changed, 36 insertions(+), 17 deletions(-) diff --git a/vgasrc/Kconfig b/vgasrc/Kconfig index 20ac2ce..00c5000 100644 --- a/vgasrc/Kconfig +++ b/vgasrc/Kconfig @@ -55,6 +55,7 @@ menu "VGA ROM" config VGA_COREBOOT depends on COREBOOT bool "coreboot linear framebuffer" + select VGA_STDVGA_FRAMEBUFFER select VGA_HIGH_FRAMEBUFFER help Build support for a vgabios wrapper around video @@ -116,7 +117,7 @@ menu "VGA ROM" Support VBE. config VGA_PCI - depends on BUILD_VGABIOS + depends on BUILD_VGABIOS && !VGA_COREBOOT bool "PCI ROM Headers" default y help diff --git a/vgasrc/cbvga.c b/vgasrc/cbvga.c index d3dc2a4..305c798 100644 --- a/vgasrc/cbvga.c +++ b/vgasrc/cbvga.c @@ -7,19 +7,19 @@ #include "biosvar.h" // GET_BDA #include "cbvga.h" // cbvga_setup #include "output.h" // dprintf +#include "stdvga.h" // SEG_CTEXT +#include "string.h" // memset16_far #include "util.h" // find_cb_table #include "vgabios.h" // VGAREG_* -#define CBMODENUM 0x140 -static struct vgamode_s CBmode VAR16 = { - MM_DIRECT, 0, 0, 0, 8, 16, 0 -}; +static int CBmode VAR16; +static struct vgamode_s CBmodeinfo VAR16; static u32 CBlinelength VAR16; struct vgamode_s *cbvga_find_mode(int mode) { - if (mode == CBMODENUM) - return &CBmode; + if (mode == GET_GLOBAL(CBmode)) + return &CBmodeinfo; return NULL; } @@ -27,7 +27,7 @@ void cbvga_list_modes(u16 seg, u16 *dest, u16 *last) { if (dest<last) { - SET_FARVAR(seg, *dest, CBMODENUM); + SET_FARVAR(seg, *dest, GET_GLOBAL(CBmode)); dest++; } SET_FARVAR(seg, *dest, 0xffff); @@ -93,7 +93,11 @@ int cbvga_set_mode(struct vgamode_s *vmode_g, int flags) { if (!(flags & MF_NOCLEARMEM)) { - int i, lines = GET_GLOBAL(CBmode.height); + if (GET_GLOBAL(CBmodeinfo.memmodel) == MM_TEXT) { + memset16_far(SEG_CTEXT, (void*)0, 0x0720, 80*25*2); + return 0; + } + int i, lines = GET_GLOBAL(CBmodeinfo.height); u32 stride = GET_GLOBAL(CBlinelength); void *dest = (void*)GET_GLOBAL(VBE_framebuffer); for (i=0; i<lines; i++, dest+=stride) @@ -137,8 +141,18 @@ cbvga_setup(void) } struct cb_framebuffer *cbfb = find_cb_subtable(cbh, CB_TAG_FRAMEBUFFER); if (!cbfb) { - dprintf(1, "Unable to find coreboot framebuffer table\n"); - return -1; + // Assume there is an EGA text framebuffer. + dprintf(1, "Did not find coreboot framebuffer - assuming EGA text\n"); + SET_VGA(CBmode, 0x03); + SET_VGA(CBlinelength, 80*2); + SET_VGA(CBmodeinfo.memmodel, MM_TEXT); + SET_VGA(CBmodeinfo.width, 80); + SET_VGA(CBmodeinfo.height, 25); + SET_VGA(CBmodeinfo.depth, 4); + SET_VGA(CBmodeinfo.cwidth, 9); + SET_VGA(CBmodeinfo.cheight, 16); + SET_VGA(CBmodeinfo.sstart, SEG_CTEXT); + return 0; } u64 addr = GET_FARVAR(0, cbfb->physical_address); @@ -155,15 +169,19 @@ cbvga_setup(void) return -1; } + SET_VGA(CBmode, 0x140); SET_VGA(VBE_framebuffer, addr); SET_VGA(VBE_total_memory, linelength * ylines); SET_VGA(CBlinelength, linelength); - SET_VGA(CBmode.width, xlines); - SET_VGA(CBmode.height, ylines); - SET_VGA(CBmode.depth, bpp); - - // Setup BDA - vga_set_mode(CBMODENUM, MF_NOCLEARMEM); + SET_VGA(CBmodeinfo.memmodel, MM_DIRECT); + SET_VGA(CBmodeinfo.width, xlines); + SET_VGA(CBmodeinfo.height, ylines); + SET_VGA(CBmodeinfo.depth, bpp); + SET_VGA(CBmodeinfo.cwidth, 8); + SET_VGA(CBmodeinfo.cheight, 16); + + // Setup BDA and clear screen. + vga_set_mode(GET_GLOBAL(CBmode), MF_NOCLEARMEM); return 0; } -- 1.8.5.3

2 1

[PATCH] vgabios: Fix PMM allocation request size.
by Kevin O'Connor 18 Feb '14

18 Feb '14

The size of a PMM memory request is in "paragraphs" so the size needs to be divided by 16. Fix the request so only the desired 512 bytes is allocated instead of 8K. Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- vgasrc/vgainit.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vgasrc/vgainit.c b/vgasrc/vgainit.c index 87bc75c..33b8eb9 100644 --- a/vgasrc/vgainit.c +++ b/vgasrc/vgainit.c @@ -75,7 +75,7 @@ allocate_extra_stack(void) "pushl %0\n" "pushw $(8|1)\n" // Permanent low memory request "pushl $0xffffffff\n" // Anonymous handle - "pushl $" __stringify(CONFIG_VGA_EXTRA_STACK_SIZE) "\n" + "pushl $" __stringify(CONFIG_VGA_EXTRA_STACK_SIZE/16) "\n" "pushw $0x00\n" // PMM allocation request "lcallw *12(%%esp)\n" "addl $16, %%esp\n" -- 1.8.5.3

1 0

Re: [SeaBIOS] [RFC][PATCH] coreboot: Add support for a "links" file to have aliases in CBFS.
by Kevin O'Connor 17 Feb '14

17 Feb '14

On Thu, Jan 30, 2014 at 12:30:11PM -0500, Kevin O'Connor wrote: > The "links" file is a newline separated list where each line is a > "link name" followed by a "destination name" separated by a space > character. For each line, SeaBIOS will consider each "link name" to > be a CBFS file that has the contents found in the CBFS file with > "destination name". > > Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> > --- > > This patch may be useful for those using intel option roms where the > same rom is used on a number of VGA devices with different > vendor/device ids. With this patch, one can make a "links" file with > the list of devices (eg, "pci8086,0a06.rom") that all point to the > same physical rom (eg, "pci8086,0406.rom"). > > It's possible to implement this aliasing approach in a number of > different ways. The approach taken in this patch is a single file > with all the link names. This was done based on feedback that there > could be 20 or more aliases for a single rom and creating individual > CBFS files for each link would be unwieldy. > > That said, I'm not sure this is the best approach. I'm circulating it > for comments. FYI, below is a version of the patch with a more resilent parser. Same comments as above. -Kevin >From 1e44227576fc76d81bfd3ee5ccb32bca827f17b1 Mon Sep 17 00:00:00 2001 Message-Id: <1e44227576fc76d81bfd3ee5ccb32bca827f17b1.1391213161.git.kevin(a)koconnor.net> From: Kevin O'Connor <kevin(a)koconnor.net> Date: Fri, 17 Jan 2014 12:11:49 -0500 Subject: [PATCH] coreboot: Add support for a "links" file to have aliases in CBFS. To: seabios(a)seabios.org The "links" file is a newline separated list where each line is a "link name" followed by a "destination name" separated by a space character. For each line, SeaBIOS will consider each "link name" to be a CBFS file that has the contents found in the CBFS file with "destination name". Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- src/fw/coreboot.c | 39 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+) diff --git a/src/fw/coreboot.c b/src/fw/coreboot.c index df8fca8..cfb7a70 100644 --- a/src/fw/coreboot.c +++ b/src/fw/coreboot.c @@ -408,6 +408,45 @@ coreboot_cbfs_init(void) fhdr = (void*)ALIGN((u32)cfile->data + cfile->rawsize , be32_to_cpu(hdr->align)); } + + // Process CBFS links file. The links file is a newline separated + // file where each line has a "link name" and a "destination name" + // separated by a space character. + char *links = romfile_loadfile("links", NULL), *next = links; + while (next) { + // Parse out linkname and destname + char *linkname = next; + next = strchr(linkname, '\n'); + if (next) + *next++ = '\0'; + while (*linkname && *linkname <= ' ') + linkname++; + char *comment = strchr(linkname, '#'); + if (comment) + *comment = '\0'; + nullTrailingSpace(linkname); + char *destname = strchr(linkname, ' '); + if (!destname) + continue; + *destname++ = '\0'; + while (*destname && *destname <= ' ') + destname++; + // Lookup destname and create new romfile entry for linkname + struct romfile_s *ufile = romfile_find(destname); + if (!ufile) + continue; + struct cbfs_romfile_s *cufile + = container_of(ufile, struct cbfs_romfile_s, file); + struct cbfs_romfile_s *cfile = malloc_tmp(sizeof(*cfile)); + if (!cfile) { + warn_noalloc(); + break; + } + memcpy(cfile, cufile, sizeof(*cfile)); + strtcpy(cfile->file.name, linkname, sizeof(cfile->file.name)); + romfile_add(&cfile->file); + } + free(links); } struct cbfs_payload_segment { -- 1.8.5.3

1 1

PC-BSD installer does not boot with 1.7.4 (bisected)
by Michael Tokarev 17 Feb '14

17 Feb '14

We have a bugreport in debian, http://bugs.debian.org/737142, stating that PC-BSD does not work with seabios-1.7.4 anymore. I digged in, and found out that it fails only with -vga std (cirrus works fine). So I bisected the issue - only changing vgabios-stdvga.bin, and found this: 9332f9b172dd59253365a83b5f1c0e40c5f6f66d is the first bad commit commit 9332f9b172dd59253365a83b5f1c0e40c5f6f66d Author: Kevin O'Connor <kevin(a)koconnor.net> Date: Sat Nov 30 12:52:44 2013 -0500 vgabios: Work around lack of support for "calll" in x86emu emulation. Replace 32 bit call instructions with 16 bit call instructions in the vgabios to workaround problems in old versions of x86emu. This change allows fc13 and fc14 to boot. (Other x86emu emulation bugs still prevent fc11 and fc12 from booting.) I'm not sure what is better - to have semi-working FC13 or non-working PC-BSD ;) Seriously, I don't really have any expirience in this area to understand what's going on. Note that PC-BSD fails clearly in some VGA-related code, and this is trivially reproducible. Thanks, /mjt

3 7

[PATCH 1/2] Add config option to move PORT_ACPI_PM_BASE out of the way
by Gerd Hoffmann 17 Feb '14

17 Feb '14

Signed-off-by: Gerd Hoffmann <kraxel(a)redhat.com> --- src/Kconfig | 7 +++++++ src/fw/paravirt.h | 9 +++++++-- 2 files changed, 14 insertions(+), 2 deletions(-) diff --git a/src/Kconfig b/src/Kconfig index cce3ad8..df3512d 100644 --- a/src/Kconfig +++ b/src/Kconfig @@ -309,6 +309,13 @@ menu "Hardware support" default y help Initialize the Memory Type Range Registers (on emulators). + config PMBASE_LOW + depends on QEMU + bool "Map ACPI PM below 0x1000" + default n + help + Map ACPI PM bar below 0x1000. This makes room in the io address + space for PCI bridges, but requires qemu 1.4+ to work correctly. config PMTIMER bool "Use ACPI timer" default y diff --git a/src/fw/paravirt.h b/src/fw/paravirt.h index 69d7248..64fe4c6 100644 --- a/src/fw/paravirt.h +++ b/src/fw/paravirt.h @@ -29,8 +29,13 @@ static inline int runningOnKVM(void) { #define PORT_SMI_STATUS 0x00b3 #define PORT_QEMU_CFG_CTL 0x0510 #define PORT_QEMU_CFG_DATA 0x0511 -#define PORT_ACPI_PM_BASE 0xb000 -#define PORT_SMB_BASE 0xb100 +#ifdef CONFIG_PMBASE_LOW +# define PORT_ACPI_PM_BASE 0x0600 +# define PORT_SMB_BASE 0x0700 +#else +# define PORT_ACPI_PM_BASE 0xb000 +# define PORT_SMB_BASE 0xb100 +#endif #define PORT_BIOS_APM 0x8900 void qemu_preinit(void); -- 1.8.3.1

2 7

[PATCH 0/4] SeaVGABIOS for native coreboot vga init
by Kevin O'Connor 12 Feb '14

12 Feb '14

This patch series enables support for writing to a framebuffer that has been setup by coreboot using its native VGA initialization. This may be a good solution for those wishing to run standard boot loaders and common operating systems without having to run a third part vgabios (on those systems where coreboot supports native vga init). -Kevin Kevin O'Connor (4): vgabios: Add option to control use of standard VGA IO ports and ranges. vgabios: Add support for writing text to framebuffers in high memory. coreboot: Make coreboot table searching available to other code. vgabios: Initial support for coreboot native vga vgabios. Makefile | 5 +- src/fw/coreboot.c | 58 ++++++++++-------- src/util.h | 3 + vgasrc/Kconfig | 26 +++++++++ vgasrc/cbvga.c | 169 +++++++++++++++++++++++++++++++++++++++++++++++++++++ vgasrc/cbvga.h | 20 +++++++ vgasrc/vgabios.c | 44 ++++++++++---- vgasrc/vgabios.h | 8 +++ vgasrc/vgafb.c | 40 +++++++++++-- vgasrc/vgahighfb.c | 154 ++++++++++++++++++++++++++++++++++++++++++++++++ vgasrc/vgahw.h | 27 +++++++++ vgasrc/vgainit.c | 3 +- 12 files changed, 514 insertions(+), 43 deletions(-) create mode 100644 vgasrc/cbvga.c create mode 100644 vgasrc/cbvga.h create mode 100644 vgasrc/vgahighfb.c -- 1.8.5.3

1 4

[PATCH] vgabios: Make sure stdvga_list_modes() doesn't overrun the buffer.
by Kevin O'Connor 12 Feb '14

12 Feb '14

Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- vgasrc/stdvgamodes.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vgasrc/stdvgamodes.c b/vgasrc/stdvgamodes.c index a97c85f..8436729 100644 --- a/vgasrc/stdvgamodes.c +++ b/vgasrc/stdvgamodes.c @@ -336,7 +336,7 @@ void stdvga_list_modes(u16 seg, u16 *dest, u16 *last) { int i; - for (i = 0; i < ARRAY_SIZE(vga_modes); i++) { + for (i = 0; i < ARRAY_SIZE(vga_modes) && dest < last; i++) { struct stdvga_mode_s *stdmode_g = &vga_modes[i]; u16 mode = GET_GLOBAL(stdmode_g->mode); if (mode == 0xffff) -- 1.8.5.3

1 0

Re: [SeaBIOS] WfW setup and DOS display.sys hanging
by Gerd Hoffmann 10 Feb '14

10 Feb '14

[ adding seabios list ] On Di, 2014-02-04 at 20:47 +0100, Kevin Wolf wrote: > Hi Gerd, > > Am 09.01.2014 um 00:20 hat Sebastian geschrieben: > > Hi, > > > > I noticed in newer versions of Qemu (Debian Jessie) a regression > > causing DISPLAY.SYS from MS-DOS 6.22 to hang the system as soon as > > loaded. > > now that you sent a pull request for a SeaBIOS update, I tried what > would happen with this one. Sadly, while it does fix my hanging Win 7 > boot, DOS still hangs while loading display.sys. [ This is http://www.kraxel.org/cgit/qemu/tag/?id=pull-roms-1 with the 1.7.4 update ] Digged out my prehistoric msdos floppy images. Can confirm the hang. Using default vga (i.e. cirrus). Switching back to lgplvgabios ("make -C roms lgplvgabios" in qemu src tree) makes the hang go away. > Any ideas where to look, or are you going to have a look yourself? When > I tried to debug it a while ago, I didn't see anything obvious, but then > I didn't invest that much time for it either. "perf kvm stat live" shows tons of NMIs: VM-EXIT Samples Samples% Time% Min Time Max Time Avg time EXCEPTION_NMI 660538 99.82% 99.40% 0us 183us 0.67us ( +- 0.14% ) EXTERNAL_INTERRUPT 1133 0.17% 0.58% 0us 43us 2.25us ( +- 4.86% ) IO_INSTRUCTION 18 0.00% 0.01% 1us 5us 3.39us ( +- 6.50% ) PENDING_INTERRUPT 17 0.00% 0.01% 1us 11us 2.15us ( +- 26.24% ) How can one get a trace log of all the kvm tracepoints these days? 'perf list' shows a bunch of kvm trace events, but "perf trace -e '*kvm*' doesn't show me anything ... cheers, Gerd > I haven't tried out any of the other cases described below. [ keeping full quote for seabios list ] > > > > The Windows for Workgroups 3.11 SETUP.EXE file hangs with a cleared > > screen and the cursor blinking in the top left of the screen. The > > DOSIDLE.EXE[1] file hangs after printing out some system > > information. > > > > Git-Bisecting returned "roms: update vgabios binaries" (commit > > 6eefccc0bb9c34051b1e21880fc3a1c1c8686edd) as the bad commit. > > > > I tested the i386-softmmu target on Debian 7.3 (wheezy), x86_64 host > > and gcc 4.7.2 (Debian 4.7.2-5) without KVM support. > > > > [1] Source and binary are available e.g. at > > http://maribu.home.xs4all.nl/zeurkous/download/mirror/dosidle/ > > > > Best Regards, > > Sebastian

3 8

Re: [SeaBIOS] [PATCH] vgabios: Simplify save/restore mechanism.
by Kevin O'Connor 10 Feb '14

10 Feb '14

On Wed, Feb 05, 2014 at 08:59:39PM -0500, Kevin O'Connor wrote: > Reorganize the save/restore functions to eliminate some boilerplate > code. FYI, I have pushed this patch and the other VGABIOS patches. -Kevin

1 0

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS