SeaBIOS

seabios@seabios.org

2 participants
3799 discussions

ps2port: `keyboard self test failed (got ab not 0xaa)`
by Paul Menzel 25 Jul '20

25 Jul '20

Dear SeaBIOS folks, On the laptop TUXEDO Book BU1406 (Clevo N240BU) with coreboot [1] and SeaBIOS master, the internal PS/2 keyboard often does not work. SeaBIOS (version rel-1.13.0-45-g6ada228) […] /7aa38000\ Start thread |7aa38000| Discarding ps2 data 41 (status=13) Searching bootorder for: HALT /7aa35000\ Start thread Found 0 lpt ports Found 0 serial ports |7aa38000| keyboard self test failed (got ab not 0xaa) \7aa38000/ End thread […] Kind regards, Paul [1]: https://review.coreboot.org/c/coreboot/+/43760/ "mb/clevo/kbl-u: Add Clevo N240BU as variant"

1 1

recent ld built for x86_64 fails to accept 16bit code as input (seabios build)
by Michael Tokarev 25 Jul '20

25 Jul '20

Hello! To clarify from the very beginning, I don't know how this part of the larger picture actually works. The end result is that seabios project, with the current source tarball available for download from https://www.seabios.org/downloads/seabios-1.13.0.tar.gz , can't be built using GNU ld as of version 2.34.90.20200706. Here's the relevant output: seabios-1.13.0 $ make silentoldconfig seabios-1.13.0 $ make V=1 PYTHON=python3 ... Linking out/rom16.o ld -T out/romlayout16.lds out/code16.o -o out/rom16.o Stripping out/rom16.strip.o strip out/rom16.o -o out/rom16.strip.o Linking out/rom.o ld -N -T out/romlayout32flat.lds out/rom16.strip.o out/rom32seg.strip.o out/code32flat.o -o out/rom.o ld: cannot use executable file 'out/rom16.strip.o' as input to a link make: *** [Makefile:187: out/rom.o] Error 1 seabios-1.13.0 $ _ Previous versions of ld accepted this file just fine and produced valid out/bios.bin in the end. I dunno if this is a problem is seabios (which has actually been already reported to seabios by Paul Menzel today), or an issue in ld. These ROMs/firmwares are a bit tricky to build.. Thanks! /mjt

10 19

[PATCH] ldnoexec: Add script to remove ET_EXEC flag from intermediate build objects
by Kevin O'Connor 24 Jul '20

24 Jul '20

Add a script to remove the ET_EXEC flag from the 16bit and "32bit segmented" intermediate objects. This avoids build failures with some linkers that will not allow linking these objects again. Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- Makefile | 9 +++++---- scripts/ldnoexec.py | 32 ++++++++++++++++++++++++++++++++ 2 files changed, 37 insertions(+), 4 deletions(-) create mode 100755 scripts/ldnoexec.py diff --git a/Makefile b/Makefile index c20be15..3d8943e 100644 --- a/Makefile +++ b/Makefile @@ -127,9 +127,10 @@ $(Q)printf '$(foreach i,$2,#include "$i"\n)' > $3.tmp.c $(Q)$(CC) -I. $1 $(CFLAGSWHOLE) -c $3.tmp.c -o $3 endef -%.strip.o: %.o +%.noexec.o: %.o @echo " Stripping $@" - $(Q)$(STRIP) $< -o $@ + $(Q)$(STRIP) $< -o $<.strip.o + $(Q)$(PYTHON) ./scripts/ldnoexec.py $<.strip.o $@ $(OUT)%.s: %.c @echo " Compiling to assembler $@" @@ -185,9 +186,9 @@ $(OUT)rom32seg.o: $(OUT)code32seg.o $(OUT)romlayout32seg.lds @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout32seg.lds $< -o $@ -$(OUT)rom.o: $(OUT)rom16.strip.o $(OUT)rom32seg.strip.o $(OUT)code32flat.o $(OUT)romlayout32flat.lds +$(OUT)rom.o: $(OUT)rom16.noexec.o $(OUT)rom32seg.noexec.o $(OUT)code32flat.o $(OUT)romlayout32flat.lds @echo " Linking $@" - $(Q)$(LD) -N -T $(OUT)romlayout32flat.lds $(OUT)rom16.strip.o $(OUT)rom32seg.strip.o $(OUT)code32flat.o -o $@ + $(Q)$(LD) -N -T $(OUT)romlayout32flat.lds $(OUT)rom16.noexec.o $(OUT)rom32seg.noexec.o $(OUT)code32flat.o -o $@ $(OUT)bios.bin.prep: $(OUT)rom.o scripts/checkrom.py @echo " Prepping $@" diff --git a/scripts/ldnoexec.py b/scripts/ldnoexec.py new file mode 100755 index 0000000..b0b32d4 --- /dev/null +++ b/scripts/ldnoexec.py @@ -0,0 +1,32 @@ +#!/usr/bin/env python2 +# Script to remove EXEC flag from an ELF file +# +# Copyright (C) 2020 Kevin O'Connor <kevin(a)koconnor.net> +# +# This file may be distributed under the terms of the GNU GPLv3 license. +import os, optparse, sys + +FLAG_OFFSET = 16 + +def main(): + # Parse command-line arguments + usage = "%prog <input_file> <output_file>" + opts = optparse.OptionParser(usage) + options, args = opts.parse_args() + if len(args) != 2: + opts.error("Incorrect number of arguments") + infilename, outfilename = args + # Read input + f = open(infilename, "rb") + srcdata = f.read() + f.close() + # Update + outdata = bytearray(srcdata) + outdata[FLAG_OFFSET] = 0x01 # change ET_EXEC to ET_REL + # Write output + f = open(outfilename, "wb") + f.write(outdata) + f.close() + +if __name__ == '__main__': + main() -- 2.26.2

2 2

[PATCH] vgabios: Fix preserve memory flag in handle_1000
by Kevin O'Connor 16 Jul '20

16 Jul '20

Commit 5108c69c made a change to the ordering of some code which broke the check for the MF_NOCLEARMEM flag. Reported by felix.von.s(a)posteo.de Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- vgasrc/vgabios.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/vgasrc/vgabios.c b/vgasrc/vgabios.c index 3b9694c..198ee55 100644 --- a/vgasrc/vgabios.c +++ b/vgasrc/vgabios.c @@ -336,6 +336,9 @@ static void handle_1000(struct bregs *regs) { int mode = regs->al & 0x7f; + int flags = MF_LEGACY | (GET_BDA(modeset_ctl) & (MF_NOPALETTE|MF_GRAYSUM)); + if (regs->al & 0x80) + flags |= MF_NOCLEARMEM; // Set regs->al if (mode > 7) @@ -345,10 +348,6 @@ handle_1000(struct bregs *regs) else regs->al = 0x30; - int flags = MF_LEGACY | (GET_BDA(modeset_ctl) & (MF_NOPALETTE|MF_GRAYSUM)); - if (regs->al & 0x80) - flags |= MF_NOCLEARMEM; - vga_set_mode(mode, flags); } -- 2.26.2

1 1

SeaVGABIOS ignores preserve-memory flag
by felix 13 Jul '20

13 Jul '20

Hello, here’s another defect I’ve found, this time in SeaVGABIOS. Video BIOS interrupt 0x10 subfunction 0 accepts a flag that determines whether video memory should be preserved after mode change. SeaVGABIOS ignores this flag. The reason is a buggy implementation of the interrupt call in vgasrc/vgabios.c: > static void > handle_1000(struct bregs *regs) > { > int mode = regs->al & 0x7f; > > // Set regs->al > if (mode > 7) > regs->al = 0x20; > else if (mode == 6) > regs->al = 0x3f; > else > regs->al = 0x30; > > int flags = MF_LEGACY | (GET_BDA(modeset_ctl) & > (MF_NOPALETTE|MF_GRAYSUM)); > if (regs->al & 0x80) > flags |= MF_NOCLEARMEM; > > vga_set_mode(mode, flags); > } The AL register’s structure member is overwritten by the chained if before the flag is checked, which means the flag is never picked up. This is a regression from commit 5108c69c47e18244206593c0c7918711311d8ef3, which removed the noclearmem variable that stashed the flag for later.[0] Sloppy refactoring aside, there was no explanation for this condition chain, so I became puzzled as to why the register was being overwritten in the first place. It seems this was meant to mimic a behaviour described in Ralf Brown’s: > Return: AL = video mode flag (Phoenix, AMI BIOS) > 20h mode > 7 > 30h modes 0-5 and 7 > 3Fh mode 6 > AL = CRT controller mode byte (Phoenix 386 BIOS v1.10) I have checked this with an Intel video BIOS on a physical machine, and it left the register untouched, so this behaviour seems hardly universal. Not even DOSBox emulates this, and I think they do have some incentives to emulate these calls as faithfully (to old hardware) as possible, so that is saying something. (Not to mention RBIL itself describes two conflicting behaviours here.) So I doubt that this condition chain needs to be kept in the first place. But even if it is kept, I think it should be moved a little later. And probably commented too. [0]: https://git.seabios.org/cgit/seabios.git/commit/vgasrc/vgabios.c?id=5108c69… — Regards, f

2 1

ld: cannot use executable file 'out/rom16.strip.o' as input to a link
by Paul Menzel 09 Jul '20

09 Jul '20

Dear SeaBIOS folks, Updating Debian Sid/unstable to binutils 2.34.90.20200706-1, the SeaBIOS for coreboot build fails with the error below. > Building ld scripts > Version: rel-1.13.0-47-g4d3f03d > Fixed space: 0xe05b-0x10000 total: 8101 slack: 9 Percent slack: 0.1% > 16bit size: 35712 > 32bit segmented size: 1665 > 32bit flat size: 32271 > 32bit flat init size: 54112 > Lowmem size: 2224 > f-segment var size: 1232 > Linking out/rom16.o > Linking out/rom32seg.o > Stripping out/rom32seg.strip.o > Stripping out/rom16.strip.o > Linking out/rom.o > ld: cannot use executable file 'out/rom16.strip.o' as input to a link > make: *** [Makefile:190: out/rom.o] Fehler 1 Kind regards, Paul

2 1

[PATCH v2] timer: Handle decrements of PIT counter
by Roman Bolshakov 26 Jun '20

26 Jun '20

2 1

[PATCH] timer: Handle decrements of PIT counter
by Roman Bolshakov 26 Jun '20

26 Jun '20

There's a fallback to PIT if TSC is not present but it doesn't work properly. It prevents boot from floppy on isapc and 486 cpu [1][2]. SeaBIOS configures PIT in Mode 2. PIT counter is decremented in the mode but timer_adjust_bits() thinks that the counter overflows and increases 32-bit tick counter on each detected "overflow". Invalid overflow detection results in 55ms time advance (1 / 18.2Hz) on each read from PIT counter. So all timers expire much faster and 5-second floppy timeout expires in 83 real microseconds (or just a bit longer). Provide counter direction to timer_adjust_bits() and normalize the counter to advance ticks in monotonically increasing TimerLast. 1. https://bugs.launchpad.net/seabios/+bug/1840719 2. https://lists.gnu.org/archive/html/qemu-devel/2019-08/msg03924.html Fixes: eac11944019 ("Unify pmtimer_read() and pittimer_read() code.") Reported-by: Philippe Mathieu-Daudé <philmd(a)redhat.com> Cc: Kevin O'Connor <kevin(a)koconnor.net> Signed-off-by: Roman Bolshakov <r.bolshakov(a)yadro.com> --- src/hw/timer.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/src/hw/timer.c b/src/hw/timer.c index 56bb289..2441402 100644 --- a/src/hw/timer.c +++ b/src/hw/timer.c @@ -156,10 +156,15 @@ u32 TimerLast VARLOW; // Add extra high bits to timers that have less than 32bits of precision. static u32 -timer_adjust_bits(u32 value, u32 validbits) +timer_adjust_bits(u32 value, u32 validbits, u8 countup) { u32 last = GET_LOW(TimerLast); - value = (last & ~validbits) | (value & validbits); + u32 validvalue; + if (countup) + validvalue = value & validbits; + else + validvalue = validbits - (value & validbits); + value = (last & ~validbits) | validvalue; if (value < last) value += validbits + 1; SET_LOW(TimerLast, value); @@ -176,11 +181,11 @@ timer_read(void) return rdtscll() >> GET_GLOBAL(ShiftTSC); if (CONFIG_PMTIMER && port != PORT_PIT_COUNTER0) // Read from PMTIMER - return timer_adjust_bits(inl(port), 0xffffff); + return timer_adjust_bits(inl(port), 0xffffff, 1); // Read from PIT. outb(PM_SEL_READBACK | PM_READ_VALUE | PM_READ_COUNTER0, PORT_PIT_MODE); u16 v = inb(PORT_PIT_COUNTER0) | (inb(PORT_PIT_COUNTER0) << 8); - return timer_adjust_bits(v, 0xffff); + return timer_adjust_bits(v, 0xffff, 0); } // Return the TSC value that is 'msecs' time in the future. -- 2.26.1

2 3

[PATCH] vga: fix cirrus bios
by Gerd Hoffmann 25 Jun '20

25 Jun '20

Little mistake, big effect. The patch adding the ati driver broke cirrus due to a missing "else", which effectively downgrades cirrus to standard vga. Fixes: 34b6ecc16074 ("vga: add atiext driver") Signed-off-by: Gerd Hoffmann <kraxel(a)redhat.com> --- vgasrc/vgahw.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vgasrc/vgahw.h b/vgasrc/vgahw.h index c774f4f2c6b7..8b64660e5ef7 100644 --- a/vgasrc/vgahw.h +++ b/vgasrc/vgahw.h @@ -36,7 +36,7 @@ static inline int vgahw_set_mode(struct vgamode_s *vmode_g, int flags) { static inline void vgahw_list_modes(u16 seg, u16 *dest, u16 *last) { if (CONFIG_VGA_CIRRUS) clext_list_modes(seg, dest, last); - if (CONFIG_VGA_ATI) + else if (CONFIG_VGA_ATI) ati_list_modes(seg, dest, last); else if (CONFIG_VGA_BOCHS) bochsvga_list_modes(seg, dest, last); -- 2.18.4

3 3

[PATCH v2] Preserve Xen DebugOutputPort
by Jason Andryuk 24 Jun '20

24 Jun '20

xen_preinit() runs early and changes DebugOutputPort. qemu_preinit() runs soon after. inb on DebugOutputPort doesn't work on Xen, so the check will always fail and DebugOutputPort will be cleared to 0 disabling output. Quick exit the function when running on Xen to preserve the modified DebugOutputPort. Signed-off-by: Jason Andryuk <jandryuk(a)gmail.com> v2 Re-word comment --- src/hw/serialio.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/hw/serialio.c b/src/hw/serialio.c index 3163344..fa663b9 100644 --- a/src/hw/serialio.c +++ b/src/hw/serialio.c @@ -106,6 +106,10 @@ u16 DebugOutputPort VARFSEG = 0x402; void qemu_debug_preinit(void) { + /* Xen doesn't support checking if debug output is active. */ + if (runningOnXen()) + return; + /* Check if the QEMU debug output port is active */ if (CONFIG_DEBUG_IO && inb(GET_GLOBAL(DebugOutputPort)) != QEMU_DEBUGCON_READBACK) -- 2.25.1

2 2

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS