SeaBIOS

seabios@seabios.org

4 participants
3801 discussions

From 16 bit to 32bit
by Daniel Castro 11 Apr '12

11 Apr '12

Hello All, Since the hypercall on 16bit is not working, how can I transition the CPU to 32bit while doing the I/O stuff. And maybe get back to 16 Bit? It appears that the 16bit CPU emulation code on Xen side can not handle the hypercall. -- +-=====---------------------------+ | +---------------------------------+ | This space intentionally blank for notetaking. | | | Daniel Castro, | | | | Consultant/Programmer.| | | | U Andes | +-------------------------------------+

1 0

Virtual Machine Crash on Hypercall
by Daniel Castro 11 Apr '12

11 Apr '12

Hello all, I am have a small problem, when running the BIOS in 16Bit if Y try to do a hypercall the virtual machines crashes. Here is the code of the hypercall: #define _hypercall2(type, name, a1, a2) \ ({ \ unsigned long __hentry = xen_hypercall_page+__HYPERVISOR_##name*32; \ long __res, __ign1, __ign2; \ asm volatile ( \ "call *%%eax" \ : "=a" (__res), "=b" (__ign1), "=c" (__ign2) \ : "0" (__hentry), "1" ((long)(a1)), "2" ((long)(a2)) \ : "memory" ); \ (type)__res; \ }) I guess the something is being passed wrong to the call, and the return address is damaged... Daniel -- +-=====---------------------------+ | +---------------------------------+ | This space intentionally blank for notetaking. | | | Daniel Castro, | | | | Consultant/Programmer.| | | | U Andes | +-------------------------------------+

1 0

Re: [SeaBIOS] Splash Screen fix (RGB bits per channel)
by Kevin O'Connor 10 Apr '12

10 Apr '12

On Tue, Apr 10, 2012 at 06:01:52AM -0600, Steve Goodrich wrote: > > On Mon, Apr 09, 2012 at 10:29:03AM -0600, Steve Goodrich wrote: > > > Anyway, the code as written was finding the first video mode of the > > correct > > > width/height and using that. Unfortunately, the mode was a 5-5-5 > > while the > > > JPG was being decoded as 5-6-5 (bits per RGB channel). > > > > What happened when the 5-5-5 mode was chosen? The code should still > > have worked (if not as pretty looking). > > The 5-5-5 mode showed the correct image, but the colors were off: for > example, some of the green areas had a significant red component (some > greens were almost entirely red). Oops - I was confusing the above with the automatic conversion from 24bpp to 16bpp. You are right that 15bpp isn't supported. It shouldn't be too hard to alter the jpeg code to support 15bpp, but it should also be safe to just skip 15bpp modes. Along the same lines, the code should really try to select a 24bpp mode if one is available. Looking closer at this, it also appears the SeaVGABIOS code isn't encoding the 15bpp mode correctly in its vbe tables. -Kevin

1 0

Re: [SeaBIOS] Splash Screen fix (RGB bits per channel)
by Kevin O'Connor 10 Apr '12

10 Apr '12

On Mon, Apr 09, 2012 at 10:29:03AM -0600, Steve Goodrich wrote: > Anyway, the code as written was finding the first video mode of the correct > width/height and using that. Unfortunately, the mode was a 5-5-5 while the > JPG was being decoded as 5-6-5 (bits per RGB channel). What happened when the 5-5-5 mode was chosen? The code should still have worked (if not as pretty looking). -Kevin

2 1

Splash Screen fix (RGB bits per channel)
by Steve Goodrich 09 Apr '12

09 Apr '12

I was trying to get a JPG splash screen working and found a problem displaying the image correctly. I'm going to bypass the arguments around including a splash screen inside a product that completes before the monitor can even sync; this is simply about code. Anyway, the code as written was finding the first video mode of the correct width/height and using that. Unfortunately, the mode was a 5-5-5 while the JPG was being decoded as 5-6-5 (bits per RGB channel). The patch below add a check for the proper bits-per-channel value before accepting/using the mode. I'm not a JPG expert, so I don't know if this will work for all situations but I thought I'd submit it for review. -- Steve G. diff --git a/src/bootsplash.c b/src/bootsplash.c index 76b72c1..69ba018 100644 --- a/src/bootsplash.c +++ b/src/bootsplash.c @@ -53,7 +53,8 @@ static int find_videomode(struct vbe_info *vesa_info, struct vbe_mode_info *mode_info , int width, int height, int bpp_req) { - dprintf(3, "Finding vesa mode with dimensions %d/%d\n", width, height); + dprintf(3, "Finding vesa mode with dimensions %d x %d (%d bpp)\n", + width, height, bpp_req); u16 *videomodes = SEGOFF_TO_FLATPTR(vesa_info->video_mode); for (;; videomodes++) { u16 videomode = *videomodes; @@ -84,6 +85,12 @@ find_videomode(struct vbe_info *vesa_info, struct vbe_mode_info *mode_info if (depth != bpp_req) continue; } + // if this is 16 bits per pixel but R:G:B is not 5:6:5 format, continue + if ((depth == 16) && + ((mode_info->red_size != 5) || + (mode_info->green_size != 6) || + (mode_info->blue_size != 5))) + continue; return videomode; } }

1 0

Re: [SeaBIOS] Version bump?
by Kevin O'Connor 09 Apr '12

09 Apr '12

On Sat, Apr 07, 2012 at 10:09:28PM +0200, Fred . wrote: > http://www.seabios.org/Releases > > Why did the version jump from 0.6.2 to 1.6.3 ? I increased the major number as I felt the project was no longer in a "beta" stage. -Kevin

2 3

Re: [SeaBIOS] Version bump?
by Peter Stuge 08 Apr '12

08 Apr '12

Fred . wrote: > Why did the version jump from 0.6.2 to 1.6.3 ? Who cares - it's just a number. //Peter

3 2

Version bump?
by Fred . 07 Apr '12

07 Apr '12

http://www.seabios.org/Releases Why did the version jump from 0.6.2 to 1.6.3 ?

1 0

Re: [SeaBIOS] Wrong address returned by container _of macro
by Kevin O'Connor 07 Apr '12

07 Apr '12

On Wed, Apr 04, 2012 at 04:04:54PM +0900, Daniel Castro wrote: > Hello All, > > I have a little problem, when I try to get my drive on boot (16bit) I > am getting the wrong address: > > This code: > struct xendrive_s * xendrive = > GLOBALFLAT2GLOBAL(container_of(GET_GLOBAL(op->drive_g), struct > xendrive_s, drive)); op is on the stack, so one must not use GET_GLOBAL on it. Also, drive_g is already a "global", so GLOBALFLAT2GLOBAL must not be called on it. So, this should read: struct xendrive_s * xendrive = container_of(op->drive_g, struct xendrive_s, drive)) If one looks at the other disk drivers, they'll see something similar - for example ahci_cmd_data() has: struct ahci_port_s *port = container_of( op->drive_g, struct ahci_port_s, drive); -Kevin

2 1

Re: [SeaBIOS] Wrong address returned by container _of macro
by Daniel Castro 06 Apr '12

06 Apr '12

On Wed, Apr 4, 2012 at 4:04 PM, Daniel Castro <evil.dani(a)gmail.com> wrote: > Hello All, > > I have a little problem, when I try to get my drive on boot (16bit) I > am getting the wrong address: > > This code: > struct xendrive_s * xendrive = > GLOBALFLAT2GLOBAL(container_of(GET_GLOBAL(op->drive_g), struct > xendrive_s, drive)); > dprintf(1,"Xendrive at:%p\n",xendrive); > Returns this: > Xendrive at:0xfff10000 > > But it should be this instead: > dprintf(1,"TEST drive 0 %p \n",i,GET_GLOBAL(xendrives[0])); > returns this: > TEST drive 0 0x000fd620 > > This last address is the address returned when done this: > struct xendrive_s *xd = malloc_fseg(sizeof(struct xendrive_s*)); > SET_FLATPTR(xendrives[count_drives],xd); > The xd pointer is later used for: > boot_add_hd(&xd->drive,desc,drives); > > Any tip to solve this will be greatly appreciated. I found a way around it, I create a global array with the drives as I add them to the boot, when the read operation is called, I compare the op->drive address to the ones I have registered, that way I can recover the struct that contains the drive and my custom struct for the drive. > > Thanks you all, > > Daniel > > > > > > -- > +-=====---------------------------+ > | +---------------------------------+ | This space intentionally blank > for notetaking. > | | | Daniel Castro, | > | | | Consultant/Programmer.| > | | | U Andes | > +-------------------------------------+ -- +-=====---------------------------+ | +---------------------------------+ | This space intentionally blank for notetaking. | | | Daniel Castro, | | | | Consultant/Programmer.| | | | U Andes | +-------------------------------------+

1 0

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS