SeaBIOS March 2010

seabios@seabios.org

16 participants
22 discussions

[PATCH] add volatile to code that gets optimized out
by Bruce Rogers 13 Mar '10

13 Mar '10

The gcc 4.5 compiler is optimizing out some code used to bring up auxillary processors. Adding volatile to the auxillary processor jump trampoline memory references gets the bios able to boot with -smp >1 again. Signed-off-by: Bruce Rogers <brogers(a)novell.com> smp.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/smp.c b/src/smp.c index dac95bf..ce7c81c 100644 --- a/src/smp.c +++ b/src/smp.c @@ -84,11 +84,11 @@ smp_probe(void) writel(&CountCPUs, 1); // Setup jump trampoline to counter code. - u64 old = *(u64*)BUILD_AP_BOOT_ADDR; + u64 old = *(volatile u64*)BUILD_AP_BOOT_ADDR; // ljmpw $SEG_BIOS, $(smp_ap_boot_code - BUILD_BIOS_ADDR) u64 new = (0xea | ((u64)SEG_BIOS<<24) | (((u32)smp_ap_boot_code - BUILD_BIOS_ADDR) << 8)); - *(u64*)BUILD_AP_BOOT_ADDR = new; + *(volatile u64*)BUILD_AP_BOOT_ADDR = new; // enable local APIC u32 val = readl(APIC_SVR); @@ -117,7 +117,7 @@ smp_probe(void) } // Restore memory. - *(u64*)BUILD_AP_BOOT_ADDR = old; + *(volatile u64*)BUILD_AP_BOOT_ADDR = old; MaxCountCPUs = qemu_cfg_get_max_cpus(); if (!MaxCountCPUs || MaxCountCPUs < CountCPUs)

2 3

USB next steps
by Kevin O'Connor 12 Mar '10

12 Mar '10

I just finished committing the EHCI support, and I thought it would be good to review what's remaining wrt USB. These are the things I'll be looking into as I get time. * refactoring the hub code. The hub and root hub code is nearly identical. * better disk timing and error support. I occasionally see USB disk boot failures that appear to be due to minor timing differences. SeaBIOS should be more resilient to these types of failures. Also, SeaBIOS should be able to reset the drive and retry on a failure. * keyboard LED support. Pressing capslock/numlock/etc. should cause the corresponding led to turn on. * disk lun support. Some devices are multi-lun - in particular, one of my multi-card reader devices shows up this way. * disk writing support. It's probably not very useful, but it's also not hard to implement. * usb mouse support. Also not very useful, but also not hard to implement. There is no OHCI support for bulk devices, and I don't plan to add it. I don't see a way to cleanly support bulk transfers from 16bit mode with OHCI. Since most people will have EHCI, I don't think it's worth adding support. Also, for the same reason, I don't plan to add keyboard led control for ohci either. -Kevin

1 0

SeaBIOS support for high-speed (EHCI) USB
by Kevin O'Connor 12 Mar '10

12 Mar '10

Hi, The latest SeaBIOS git now has support for EHCI USB controllers. This is in addition to the UHCI and OHCI support already in SeaBIOS. To obtain SeaBIOS, see: http://seabios.org/Download I've tested the EHCI support on my epia-cn. Booting from USB drives is now noticeably faster. Currently SeaBIOS supports USB keyboards, USB hubs, and USB drives. Help with testing on a wider range of hardware would be appreciated. -Kevin

1 0

Re: [SeaBIOS] [Qemu-devel] Re: Need a way disable gPXE boot
by Kevin O'Connor 10 Mar '10

10 Mar '10

On Tue, Mar 09, 2010 at 02:11:20PM +0000, Richard W.M. Jones wrote: > I noticed also with libguestfs that boot times have spiralled out of > control again, and it seems to be down to the SeaBIOS change. SeaBIOS > accounts for _9_ seconds of the boot sequence, up from very small > (fraction of a second) for the old Bochs BIOS[1]. I'm surprised - a significant amount of effort has gone into making SeaBIOS boot quickly. On my machine, the OS starts loading in well under a second. I'm not familiar with libguestfs - is that doing something to alter the hardware emulation? -Kevin

2 1

[PATCH 0/5] Add USB EHCI controller support.
by Kevin O'Connor 10 Mar '10

10 Mar '10

This patch set adds support for high-speed USB (EHCI) controllers. I've tested this on my epia-cn machine. The EHCI support noticeably decreases the time required to boot from a USB drive. This support includes the code necessary to hand off full/low speed devices to "companion controllers" (both uhci and ohci). It also supports full/low speed devices attached to high-speed hubs. This is an initial implementation, but I think it's ready for more wide spread testing. Feedback is welcome. -Kevin Kevin O'Connor (5): Dynamically allocate USB controller structures. Reduce size of USB 'struct uhci_td'. Minor - USB OHCI interrupt queue should be one larger. Some USB UHCI and OHCI fixes and cleanups. Add USB EHCI controller support. Makefile | 2 +- src/config.h | 2 + src/usb-ehci.c | 751 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ src/usb-ehci.h | 173 +++++++++++++ src/usb-hub.c | 3 +- src/usb-hub.h | 2 + src/usb-ohci.c | 232 ++++++++++-------- src/usb-ohci.h | 3 +- src/usb-uhci.c | 214 +++++++++------- src/usb-uhci.h | 5 +- src/usb.c | 82 +++++-- src/usb.h | 31 +-- 12 files changed, 1261 insertions(+), 239 deletions(-) create mode 100644 src/usb-ehci.c create mode 100644 src/usb-ehci.h

1 5

[PATCH 0/4] Further parallelize USB init by launching a thread per usb port.
by Kevin O'Connor 10 Mar '10

10 Mar '10

Main work is in last patch - first three patches are just cleanups. This patch series improves the boot time on my epia-cn machine, and it also sets up for further ehancements that will follow. -Kevin Kevin O'Connor (4): Prefer passing a USB "pipe" structure over a USB endp encoding. Only compile usb-hub.c and paravirt.c with 32bit code. Introduce simple "mutex" locking code. Further parallelize USB init by launching a thread per usb port. Makefile | 4 +- src/ps2port.c | 2 +- src/stacks.c | 20 ++++ src/usb-hid.c | 25 +++-- src/usb-hid.h | 5 +- src/usb-hub.c | 190 +++++++++++++++++++++------------ src/usb-hub.h | 13 ++- src/usb-msc.c | 9 +- src/usb-msc.h | 4 +- src/usb-ohci.c | 325 +++++++++++++++++++++++++++++++++++++++---------------- src/usb-ohci.h | 6 +- src/usb-uhci.c | 319 ++++++++++++++++++++++++++++++++++++++----------------- src/usb-uhci.h | 12 +-- src/usb.c | 167 +++++++++++++++++++---------- src/usb.h | 24 +++-- src/util.h | 3 + 16 files changed, 767 insertions(+), 361 deletions(-)

1 4

Re: [SeaBIOS] [PATCH 2/4] KVM: Rework VCPU state writeback API
by Kevin O'Connor 09 Mar '10

09 Mar '10

On Thu, Mar 04, 2010 at 01:21:12AM -0300, Marcelo Tosatti wrote: > The regression seems to be caused by seabios commit d7e998f. Kevin, the > failure can be seen on the attached screenshot, which happens on the > first reboot of WinXP 32 installation (after copying files etc). Sorry - I also noticed a bug in that commit recently. I pushed the fix I had in my local tree. -Kevin

3 4

too big seabios (1MB)
by Rudolf Marek 09 Mar '10

09 Mar '10

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello, With debian unstable binutils 2.18 and 2.20 I always get quite big bios.bin.elf Image. I believe it is because rom.o has alignes for 2**21 Yes this is default align on AMD64. out/rom.o: file format elf32-i386 out/rom.o architecture: i386, flags 0x00000112: EXEC_P, HAS_SYMS, D_PAGED start address 0x000fdf9d Program Header: LOAD off 0x000ef000 vaddr 0x000ef000 paddr 0x000ef000 align 2**21 filesz 0x00011000 memsz 0x00011000 flags r-x Not sure how to fix it? I tried adding . = ALIGN(4096) to the linker script but it complains about truncated relocs. Rudolf -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkuSY+8ACgkQ3J9wPJqZRNUGsACcCOaux4C6Y3MEtPR+qe9sVS9A h+MAoLn33J7XzcZUrPq5Dif2X5MxSl4h =I7RK -----END PGP SIGNATURE-----

3 3

some USB/VGA success story
by Rudolf Marek 06 Mar '10

06 Mar '10

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi all, I managed to compile seabios on 32 bit host. I tested it on following hw: Asus M2V-MX SE (K8M890/VT8237S) + coreboot Bus 005 Device 002: ID 1e3d:6025 Bus 002 Device 004: ID 04f2:0116 Chicony Electronics Co., Ltd KU-2971 German Keyboard Bus 002 Device 005: ID 0430:0005 Sun Microsystems, Inc. Type 6 Keyboard It has UHCI and EHCI the USB stick was EHCI connected. In summary, USB keyboard works! Both of them, autorepeat is working too. The only minor thing I could not use them BOTH connected ;) USB booting works too! I booted from USB stick the FreeDOS, USB keyboard still worked :) !!! And to make it even more fancy, I used the vgabios.bin from the SeaBIOS. (marked in CBFS as pci1106,3230.rom) Because Luc did the VGA init in the Coreboot of the Chrome9 VGA, the legacy VGA does work with legacy SeaBIOS option rom ;) There is some glitch that part of the image contain old data during some editing experiments in grub, but maybe just page switching is not correct? Dont know. But it was usable :) Rudolf -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkuSbYgACgkQ3J9wPJqZRNUVJwCgvBnMpAlhosXZ7F2TSL9Ff6yp c80AoJv1evpcd/zMc70NZgteDwX3Zo1C =HHir -----END PGP SIGNATURE-----

1 0

Re: [SeaBIOS] [Qemu-devel] SeaBIOS error with Juniper FreeBSD kernel
by Kevin O'Connor 04 Mar '10

04 Mar '10

I'm CC'ing the seabios list. On Wed, Mar 03, 2010 at 12:05:52PM +0200, Gleb Natapov wrote: > On Mon, Mar 01, 2010 at 07:09:56PM -0500, Kevin O'Connor wrote: > > On Mon, Mar 01, 2010 at 05:36:24PM +0800, Keen Shoes wrote: > > > Hi, Gleb > > > The tested bios, both 0.5.0 and 0.5.1 was downloaded from > > > here.http://linuxtogo.org/~kevin/SeaBIOS/. > > > And the 0.5.0 works well, while bios.0.5.1 failed to boot the > > > same JunOS images. By my experience, the same problem occured on > > > most of JunOS version, tested on 8.5, 9.3 and 9.6. > > > > The 0.5.0 release is very recent, I don't see any significant changes > > in smbios since that time. Going through the bisect process would > > help a lot here. > > > Bisect points to fb214dc70aaeb7b4bbaf270bb9566843a0606933 and I > confirmed manually that one commit before that works, _but_ > reverting this commit from master doesn't help. This indicates to me there is an indirect failure occuring -- something reading memory that it shouldn't. I'd guess random changes in the build are tickling the behavior. It would exmplain why Anthony's 0.5.0 fails while my 0.5.0 is okay, and why disabling SMBIOS fixes it for Brandon. -Kevin

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS March 2010