SeaBIOS November 2013

seabios@seabios.org

18 participants
43 discussions

[PATCH] vgasrc: add HDTV resolutions (1280x768, 1280x720, 1920x1080)
by Michael Tokarev 30 Nov '13

30 Nov '13

The same set were added to vgabios at version 0.7a. Signed-off-by: Michael Tokarev <mjt(a)tls.msk.ru> --- vgasrc/bochsvga.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/vgasrc/bochsvga.c b/vgasrc/bochsvga.c index b0ba1ec..3c070c6 100644 --- a/vgasrc/bochsvga.c +++ b/vgasrc/bochsvga.c @@ -68,6 +68,9 @@ static struct bochsvga_mode { 0x14a, { MM_DIRECT, 1152, 864, 16, 8, 16, SEG_GRAPH } }, { 0x14b, { MM_DIRECT, 1152, 864, 24, 8, 16, SEG_GRAPH } }, { 0x14c, { MM_DIRECT, 1152, 864, 32, 8, 16, SEG_GRAPH } }, + { 0x175, { MM_DIRECT, 1280, 768, 16, 8, 16, SEG_GRAPH } }, + { 0x176, { MM_DIRECT, 1280, 768, 24, 8, 16, SEG_GRAPH } }, + { 0x177, { MM_DIRECT, 1280, 768, 32, 8, 16, SEG_GRAPH } }, { 0x178, { MM_DIRECT, 1280, 800, 16, 8, 16, SEG_GRAPH } }, { 0x179, { MM_DIRECT, 1280, 800, 24, 8, 16, SEG_GRAPH } }, { 0x17a, { MM_DIRECT, 1280, 800, 32, 8, 16, SEG_GRAPH } }, @@ -89,6 +92,14 @@ static struct bochsvga_mode { 0x18a, { MM_DIRECT, 2560, 1600, 16, 8, 16, SEG_GRAPH } }, { 0x18b, { MM_DIRECT, 2560, 1600, 24, 8, 16, SEG_GRAPH } }, { 0x18c, { MM_DIRECT, 2560, 1600, 32, 8, 16, SEG_GRAPH } }, + { 0x18d, { MM_DIRECT, 1280, 720, 16, 8, 16, SEG_GRAPH } }, + { 0x18d, { MM_DIRECT, 1280, 720, 24, 8, 16, SEG_GRAPH } }, + { 0x18e, { MM_DIRECT, 1280, 720, 32, 8, 16, SEG_GRAPH } }, + { 0x18f, { MM_DIRECT, 1280, 720, 16, 8, 16, SEG_GRAPH } }, + { 0x190, { MM_DIRECT, 1920, 1080, 16 8, 16, SEG_GRAPH } }, + { 0x191, { MM_DIRECT, 1920, 1080, 24 8, 16, SEG_GRAPH } }, + { 0x192, { MM_DIRECT, 1920, 1080, 32, 8, 16, SEG_GRAPH } }, + }; static int is_bochsvga_mode(struct vgamode_s *vmode_g) -- 1.7.10.4

3 4

Re: [SeaBIOS] [PATCH] vgasrc: add HDTV resolutions (1280x768, 1280x720, 1920x1080)
by Michael Tokarev 30 Nov '13

30 Nov '13

30.11.2013 04:17, Fred . wrote: > 1920 × 1200 is a pretty common resolution. > On computers, not on TVs. I wanted those to be the same as in lgpl vgabios, the patch that added those to vgabios was named like this. Yes, this resolution is now common on computers, but it come from TVs - it is the standard full-HD resolution. > Also, now there is UHD TV resolutions. The 4K resolutions. > Like 3840 × 2160 and 4096 × 2160. That should be a new patch, I think... ;) /mjt

1 0

seavgabios resolutions, win8
by Michael Tokarev 29 Nov '13

29 Nov '13

I tried switching from plex/bochs vgabios to seavgabios, and made this switch for qemu-1.7 package in debian, at least to see how it goes. And almost immediately got a complain that with new qemu (which uses seavgabios not vgabios), windows VMs don't support 1920x1080 resolutions anymore. Almost immediately we cooked a patch which restored the mode(s) in question, here it is: --- a/vgasrc/bochsvga.c +++ b/vgasrc/bochsvga.c @@ -89,6 +89,9 @@ { 0x18a, { MM_DIRECT, 2560, 1600, 16, 8, 16, SEG_GRAPH } }, { 0x18b, { MM_DIRECT, 2560, 1600, 24, 8, 16, SEG_GRAPH } }, { 0x18c, { MM_DIRECT, 2560, 1600, 32, 8, 16, SEG_GRAPH } }, + { 0x18d, { MM_DIRECT, 1920, 1080, 16, 8, 16, SEG_GRAPH } }, + { 0x18e, { MM_DIRECT, 1920, 1080, 24, 8, 16, SEG_GRAPH } }, + { 0x18f, { MM_DIRECT, 1920, 1080, 32, 8, 16, SEG_GRAPH } }, }; static int is_bochsvga_mode(struct vgamode_s *vmode_g) Should something like this be applied? Of the same theme, there's a wishlist bugreport against vgabios in Debian, to add a few more resolutions, http://bugs.debian.org/669114 -- it adds 2048x1536 resolution, which is also missing in seavgabios. Also, vgabios 0.7 provided a few more resolutions, like 1280x768 and 1280x720. I'm not sure I understand how list of modes is maintained/used in seavgabios. The above patch at least allows one of the very common modes to be selected in windows clients. And while at it, there's another bugreport against vgabios package in Debian, see http://bugs.debian.org/685097 . It looks like windows8 isn't happy with vgabios-provided modes and does not let to change resolution at all, sticking at 1024x768. That bugreport points to a patchset, http://lists.xen.org/archives/html/xen-devel/2012-05/msg00260.html (this is just one patch from it), which has been applied to xen copy of vgabios to solve exactly this issue with win8, and reportedly it works there. It looks like something similar should be done in seavgabios too, since with it, win8 guest is also stuck at the same resolution. Thanks, /mjt

3 6

[PATCH] Move low-level hardware writing from output.c to new file hw/serialio.c.
by Kevin O'Connor 29 Nov '13

29 Nov '13

Avoid hardware specific code in output.c. This will reduce the amount of change needed to output.c as support for more serial hardware is added. This patch also renames some functions to improve the naming scheme. Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- Makefile | 2 +- src/boot.c | 4 +-- src/fw/coreboot.c | 2 +- src/fw/xen.c | 1 + src/hw/serialio.c | 89 +++++++++++++++++++++++++++++++++++++++++++++++ src/hw/serialio.h | 8 +++++ src/optionroms.c | 2 +- src/output.c | 101 ++++++++++++++++-------------------------------------- src/output.h | 3 +- src/post.c | 2 +- src/resume.c | 2 +- src/util.h | 2 +- 12 files changed, 137 insertions(+), 81 deletions(-) create mode 100644 src/hw/serialio.c diff --git a/Makefile b/Makefile index 855fdf4..3aefbe6 100644 --- a/Makefile +++ b/Makefile @@ -29,7 +29,7 @@ IASL:=iasl SRCBOTH=misc.c stacks.c output.c string.c x86.c block.c cdrom.c mouse.c kbd.c \ serial.c clock.c resume.c pnpbios.c vgahooks.c pcibios.c apm.c \ fw/smp.c \ - hw/pci.c hw/timer.c hw/rtc.c hw/dma.c hw/pic.c hw/ps2port.c \ + hw/pci.c hw/timer.c hw/rtc.c hw/dma.c hw/pic.c hw/ps2port.c hw/serialio.c \ hw/usb.c hw/usb-uhci.c hw/usb-ohci.c hw/usb-ehci.c hw/usb-xhci.c \ hw/usb-hid.c hw/usb-msc.c hw/usb-uas.c \ hw/blockcmd.c hw/floppy.c hw/ata.c hw/ramdisk.c \ diff --git a/src/boot.c b/src/boot.c index b14ced9..1c74d57 100644 --- a/src/boot.c +++ b/src/boot.c @@ -743,7 +743,7 @@ int BootSequence VARLOW = -1; void VISIBLE32FLAT handle_18(void) { - debug_serial_preinit(); + debug_preinit(); debug_enter(NULL, DEBUG_HDL_18); int seq = BootSequence + 1; BootSequence = seq; @@ -754,7 +754,7 @@ handle_18(void) void VISIBLE32FLAT handle_19(void) { - debug_serial_preinit(); + debug_preinit(); debug_enter(NULL, DEBUG_HDL_19); BootSequence = 0; do_boot(0); diff --git a/src/fw/coreboot.c b/src/fw/coreboot.c index 70a99a0..5b7e77a 100644 --- a/src/fw/coreboot.c +++ b/src/fw/coreboot.c @@ -202,7 +202,7 @@ fail: return; } -void debug_cbmem(char c) +void coreboot_debug_putc(char c) { if (!CONFIG_DEBUG_COREBOOT) return; diff --git a/src/fw/xen.c b/src/fw/xen.c index 9b00030..d5f5ed3 100644 --- a/src/fw/xen.c +++ b/src/fw/xen.c @@ -5,6 +5,7 @@ // This file may be distributed under the terms of the GNU LGPLv3 license. #include "config.h" +#include "hw/serialio.h" // DebugOutputPort #include "malloc.h" // memalign_high #include "memmap.h" // add_e820 #include "output.h" // dprintf diff --git a/src/hw/serialio.c b/src/hw/serialio.c new file mode 100644 index 0000000..6486fc0 --- /dev/null +++ b/src/hw/serialio.c @@ -0,0 +1,89 @@ +// Low-level serial (and serial-like) device access. +// +// Copyright (C) 2008-1013 Kevin O'Connor <kevin(a)koconnor.net> +// +// This file may be distributed under the terms of the GNU LGPLv3 license. + +#include "config.h" // CONFIG_DEBUG_SERIAL +#include "fw/paravirt.h" // RunningOnQEMU +#include "output.h" // dprintf +#include "serialio.h" // serial_debug_preinit +#include "x86.h" // outb + + +/**************************************************************** + * Serial port debug output + ****************************************************************/ + +#define DEBUG_TIMEOUT 100000 + +// Setup the debug serial port for output. +void +serial_debug_preinit(void) +{ + if (!CONFIG_DEBUG_SERIAL) + return; + // setup for serial logging: 8N1 + u8 oldparam, newparam = 0x03; + oldparam = inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_LCR); + outb(newparam, CONFIG_DEBUG_SERIAL_PORT+SEROFF_LCR); + // Disable irqs + u8 oldier, newier = 0; + oldier = inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_IER); + outb(newier, CONFIG_DEBUG_SERIAL_PORT+SEROFF_IER); + + if (oldparam != newparam || oldier != newier) + dprintf(1, "Changing serial settings was %x/%x now %x/%x\n" + , oldparam, oldier, newparam, newier); +} + +// Write a character to the serial port. +static void +serial_debug(char c) +{ + if (!CONFIG_DEBUG_SERIAL) + return; + int timeout = DEBUG_TIMEOUT; + while ((inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_LSR) & 0x20) != 0x20) + if (!timeout--) + // Ran out of time. + return; + outb(c, CONFIG_DEBUG_SERIAL_PORT+SEROFF_DATA); +} + +void +serial_debug_putc(char c) +{ + if (c == '\n') + serial_debug('\r'); + serial_debug(c); +} + +// Make sure all serial port writes have been completely sent. +void +serial_debug_flush(void) +{ + if (!CONFIG_DEBUG_SERIAL) + return; + int timeout = DEBUG_TIMEOUT; + while ((inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_LSR) & 0x60) != 0x60) + if (!timeout--) + // Ran out of time. + return; +} + + +/**************************************************************** + * QEMU debug port + ****************************************************************/ + +u16 DebugOutputPort VARFSEG = 0x402; + +// Write a character to the special debugging port. +void +qemu_debug_putc(char c) +{ + if (CONFIG_DEBUG_IO && runningOnQEMU()) + // Send character to debug port. + outb(c, GET_GLOBAL(DebugOutputPort)); +} diff --git a/src/hw/serialio.h b/src/hw/serialio.h index 883fc9c..88296fe 100644 --- a/src/hw/serialio.h +++ b/src/hw/serialio.h @@ -1,6 +1,8 @@ #ifndef __SERIALIO_H #define __SERIALIO_H +#include "types.h" // u16 + #define PORT_LPT2 0x0278 #define PORT_SERIAL4 0x02e8 #define PORT_SERIAL2 0x02f8 @@ -18,4 +20,10 @@ #define SEROFF_LSR 5 #define SEROFF_MSR 6 +void serial_debug_preinit(void); +void serial_debug_putc(char c); +void serial_debug_flush(void); +extern u16 DebugOutputPort; +void qemu_debug_putc(char c); + #endif // serialio.h diff --git a/src/optionroms.c b/src/optionroms.c index d7c8aef..1ff73c4 100644 --- a/src/optionroms.c +++ b/src/optionroms.c @@ -45,7 +45,7 @@ __callrom(struct rom_header *rom, u16 offset, u16 bdf) farcall16big(&br); finish_preempt(); - debug_serial_preinit(); + debug_preinit(); } // Execute a given option rom at the standard entry vector. diff --git a/src/output.c b/src/output.c index b47625f..1d882fa 100644 --- a/src/output.c +++ b/src/output.c @@ -1,6 +1,6 @@ // Raw screen writing and debug output code. // -// Copyright (C) 2008,2009 Kevin O'Connor <kevin(a)koconnor.net> +// Copyright (C) 2008-2013 Kevin O'Connor <kevin(a)koconnor.net> // // This file may be distributed under the terms of the GNU LGPLv3 license. @@ -10,8 +10,7 @@ #include "bregs.h" // struct bregs #include "config.h" // CONFIG_* #include "biosvar.h" // GET_GLOBAL -#include "fw/paravirt.h" // PlatformRunningOn -#include "hw/serialio.h" // SEROFF_IER +#include "hw/serialio.h" // serial_debug_putc #include "malloc.h" // malloc_tmp #include "output.h" // dprintf #include "stacks.h" // call16_int @@ -27,81 +26,41 @@ struct putcinfo { * Debug output ****************************************************************/ -#define DEBUG_TIMEOUT 100000 - -u16 DebugOutputPort VARFSEG = 0x402; - +// Setup debugging port(s). void -debug_serial_preinit(void) +debug_preinit(void) { - if (!CONFIG_DEBUG_SERIAL) - return; - // setup for serial logging: 8N1 - u8 oldparam, newparam = 0x03; - oldparam = inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_LCR); - outb(newparam, CONFIG_DEBUG_SERIAL_PORT+SEROFF_LCR); - // Disable irqs - u8 oldier, newier = 0; - oldier = inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_IER); - outb(newier, CONFIG_DEBUG_SERIAL_PORT+SEROFF_IER); - - if (oldparam != newparam || oldier != newier) - dprintf(1, "Changing serial settings was %x/%x now %x/%x\n" - , oldparam, oldier, newparam, newier); -} - -// Write a character to the serial port. -static void -debug_serial(char c) -{ - if (!CONFIG_DEBUG_SERIAL) - return; - int timeout = DEBUG_TIMEOUT; - while ((inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_LSR) & 0x20) != 0x20) - if (!timeout--) - // Ran out of time. - return; - outb(c, CONFIG_DEBUG_SERIAL_PORT+SEROFF_DATA); + serial_debug_preinit(); } -// Make sure all serial port writes have been completely sent. +// Write a character to debug port(s). static void -debug_serial_flush(void) +debug_putc(struct putcinfo *action, char c) { - if (!CONFIG_DEBUG_SERIAL) + if (! CONFIG_DEBUG_LEVEL) return; - int timeout = DEBUG_TIMEOUT; - while ((inb(CONFIG_DEBUG_SERIAL_PORT+SEROFF_LSR) & 0x60) != 0x60) - if (!timeout--) - // Ran out of time. - return; + qemu_debug_putc(c); + if (!MODESEGMENT) + coreboot_debug_putc(c); + serial_debug_putc(c); } -// Write a character to debug port(s). +// Flush any pending output to debug port(s). static void -putc_debug(struct putcinfo *action, char c) +debug_flush(void) { - if (! CONFIG_DEBUG_LEVEL) - return; - if (CONFIG_DEBUG_IO && runningOnQEMU()) - // Send character to debug port. - outb(c, GET_GLOBAL(DebugOutputPort)); - if (!MODESEGMENT) - debug_cbmem(c); - if (c == '\n') - debug_serial('\r'); - debug_serial(c); + serial_debug_flush(); } -// In segmented mode just need a dummy variable (putc_debug is always +// In segmented mode just need a dummy variable (debug_putc is always // used anyway), and in 32bit flat mode need a pointer to the 32bit -// instance of putc_debug(). +// instance of debug_putc(). #if MODE16 static struct putcinfo debuginfo VAR16; #elif MODESEGMENT static struct putcinfo debuginfo VAR32SEG; #else -static struct putcinfo debuginfo = { putc_debug }; +static struct putcinfo debuginfo = { debug_putc }; #endif @@ -123,16 +82,16 @@ screenc(char c) // Handle a character from a printf request. static void -putc_screen(struct putcinfo *action, char c) +screen_putc(struct putcinfo *action, char c) { if (ScreenAndDebug) - putc_debug(&debuginfo, c); + debug_putc(&debuginfo, c); if (c == '\n') screenc('\r'); screenc(c); } -static struct putcinfo screeninfo = { putc_screen }; +static struct putcinfo screeninfo = { screen_putc }; /**************************************************************** @@ -145,7 +104,7 @@ putc(struct putcinfo *action, char c) { if (MODESEGMENT) { // Only debugging output supported in segmented mode. - putc_debug(action, c); + debug_putc(action, c); return; } @@ -348,7 +307,7 @@ panic(const char *fmt, ...) va_start(args, fmt); bvprintf(&debuginfo, fmt, args); va_end(args); - debug_serial_flush(); + debug_flush(); } // XXX - use PANIC PORT. @@ -365,10 +324,10 @@ __dprintf(const char *fmt, ...) struct thread_info *cur = getCurThread(); if (cur != &MainThread) { // Show "thread id" for this debug message. - putc_debug(&debuginfo, '|'); + debug_putc(&debuginfo, '|'); puthex(&debuginfo, (u32)cur, 8); - putc_debug(&debuginfo, '|'); - putc_debug(&debuginfo, ' '); + debug_putc(&debuginfo, '|'); + debug_putc(&debuginfo, ' '); } } @@ -376,7 +335,7 @@ __dprintf(const char *fmt, ...) va_start(args, fmt); bvprintf(&debuginfo, fmt, args); va_end(args); - debug_serial_flush(); + debug_flush(); } void @@ -388,7 +347,7 @@ printf(const char *fmt, ...) bvprintf(&screeninfo, fmt, args); va_end(args); if (ScreenAndDebug) - debug_serial_flush(); + debug_flush(); } @@ -479,7 +438,7 @@ hexdump(const void *d, int len) d+=4; } putc(&debuginfo, '\n'); - debug_serial_flush(); + debug_flush(); } static void @@ -503,7 +462,7 @@ __debug_isr(const char *fname) { puts_cs(&debuginfo, fname); putc(&debuginfo, '\n'); - debug_serial_flush(); + debug_flush(); } // Function called on handler startup. diff --git a/src/output.h b/src/output.h index bdfdd63..a13d55b 100644 --- a/src/output.h +++ b/src/output.h @@ -4,8 +4,7 @@ #include "types.h" // u32 // output.c -extern u16 DebugOutputPort; -void debug_serial_preinit(void); +void debug_preinit(void); void panic(const char *fmt, ...) __attribute__ ((format (printf, 1, 2))) __noreturn; void printf(const char *fmt, ...) diff --git a/src/post.c b/src/post.c index b634686..ace6f4c 100644 --- a/src/post.c +++ b/src/post.c @@ -317,7 +317,7 @@ handle_post(void) if (!CONFIG_QEMU && !CONFIG_COREBOOT) return; - debug_serial_preinit(); + debug_preinit(); dprintf(1, "Start bios (version %s)\n", VERSION); // Check if we are running under Xen. diff --git a/src/resume.c b/src/resume.c index fc2fee9..d69429c 100644 --- a/src/resume.c +++ b/src/resume.c @@ -25,7 +25,7 @@ void VISIBLE16 handle_resume(void) { ASSERT16(); - debug_serial_preinit(); + debug_preinit(); int status = rtc_read(CMOS_RESET_CODE); rtc_write(CMOS_RESET_CODE, 0); dprintf(1, "In resume (status=%d)\n", status); diff --git a/src/util.h b/src/util.h index e6a6cb5..1b7d525 100644 --- a/src/util.h +++ b/src/util.h @@ -78,7 +78,7 @@ void *find_acpi_rsdp(void); // fw/coreboot.c extern const char *CBvendor, *CBpart; struct cbfs_file; -void debug_cbmem(char c); +void coreboot_debug_putc(char c); void cbfs_run_payload(struct cbfs_file *file); void coreboot_platform_setup(void); void cbfs_payload_setup(void); -- 1.8.3.1

1 0

Re: [SeaBIOS] [PATCH v3] i386: Add _PXM ACPI method to CPU objects
by Igor Mammedov 29 Nov '13

29 Nov '13

On Wed, 27 Nov 2013 19:38:44 +0100 Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> wrote: > This patch adds a _PXM method to ACPI CPU objects for the pc machine. The _PXM > value is derived from the passed in guest info, same way as CPU SRAT entries. > > Currently, CPU SRAT entries are only enabled for cpus that are already present > in the system. The SRAT entries for hotpluggable processors are disabled (flags > bit 0 set to 0 in hw/i385/acpi-build.c:build_srat). Section 5.2.16.1 of ACPI > spec mentions "If the Local APIC ID of a dynamically added processor is not > present in the SRAT, a _PXM object must exist for the processor’s device or one > of its ancestors in the ACPI Namespace." Since SRAT entries are not available > for the hot-pluggable processors, a _PXM method must exist for them. Otherwise, > the CPU is hot-added in the wrong NUMA node (default node 0). > > Even if CPU SRAT entries are enabled, _PXM method is what the linux kernel > consults on hot-add time. Section 17.2.1 of ACPI spec mentions " OSPM will > consume the SRAT only at boot time. OSPM should use _PXM for any devices that > are hot-added into the system after boot up." To be more precise if SRAT > information is available to the guest kernel, it is used. However, parsed SRAT > info is reset and lost after hot-remove operations, see kernel commit c4c60524. > This means that on a hot-unplug / hot-replug scenario, and without a _PXM > method, the kernel may put a CPU on different nodes because SRAT info has been > reset by a previous hot-remove operation. > > The above hot-remove/hot-add scenario has been tested on master, plus cpu-del > patches from: > https://lists.gnu.org/archive/html/qemu-devel/2013-10/msg01085.html > With the curret _PXM patch, hot-added CPUs are always placed into the correct > NUMA node, regardless of kernel behaviour. > > v1->v2: > Make method return a DWORD integer > Tested on qemu master + cpu-del patches > v2->v3: > Add changed hw/i386/sdt-proc.hex.generated file > Change PXM constant name to CPXM > > Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> > Reviewed-by: Thilo Fromm <t-lo(a)thilo-fromm.de> > > --- > hw/i386/acpi-build.c | 5 ++++ > hw/i386/ssdt-proc.dsl | 5 ++++ > hw/i386/ssdt-proc.hex.generated | 57 ++++++++++++++++++++++++++++----------- > 3 files changed, 51 insertions(+), 16 deletions(-) > > diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c > index b48c930..387a869 100644 > --- a/hw/i386/acpi-build.c > +++ b/hw/i386/acpi-build.c > @@ -605,6 +605,7 @@ static inline char acpi_get_hex(uint32_t val) > #define ACPI_PROC_OFFSET_CPUHEX (*ssdt_proc_name - *ssdt_proc_start + 2) > #define ACPI_PROC_OFFSET_CPUID1 (*ssdt_proc_name - *ssdt_proc_start + 4) > #define ACPI_PROC_OFFSET_CPUID2 (*ssdt_proc_id - *ssdt_proc_start) > +#define ACPI_PROC_OFFSET_CPUPXM (*ssdt_proc_pxm - *ssdt_proc_start) > #define ACPI_PROC_SIZEOF (*ssdt_proc_end - *ssdt_proc_start) > #define ACPI_PROC_AML (ssdp_proc_aml + *ssdt_proc_start) > > @@ -726,6 +727,10 @@ build_ssdt(GArray *table_data, GArray *linker, > proc[ACPI_PROC_OFFSET_CPUHEX+1] = acpi_get_hex(i); > proc[ACPI_PROC_OFFSET_CPUID1] = i; > proc[ACPI_PROC_OFFSET_CPUID2] = i; > + proc[ACPI_PROC_OFFSET_CPUPXM] = guest_info->node_cpu[i]; > + proc[ACPI_PROC_OFFSET_CPUPXM + 1] = 0; > + proc[ACPI_PROC_OFFSET_CPUPXM + 2] = 0; > + proc[ACPI_PROC_OFFSET_CPUPXM + 3] = 0; > } > > /* build this code: > diff --git a/hw/i386/ssdt-proc.dsl b/hw/i386/ssdt-proc.dsl > index 8229bfd..52b44e3 100644 > --- a/hw/i386/ssdt-proc.dsl > +++ b/hw/i386/ssdt-proc.dsl > @@ -47,6 +47,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) > * also updating the C code. > */ > Name(_HID, "ACPI0007") > + ACPI_EXTRACT_NAME_DWORD_CONST ssdt_proc_pxm > + Name(CPXM, 0xAAAAAAAA) > External(CPMA, MethodObj) > External(CPST, MethodObj) > External(CPEJ, MethodObj) > @@ -59,5 +61,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) > Method(_EJ0, 1, NotSerialized) { > CPEJ(ID, Arg0) > } > + Method(_PXM, 0) { > + Return (CPXM) > + } > } > } > diff --git a/hw/i386/ssdt-proc.hex.generated b/hw/i386/ssdt-proc.hex.generated > index bb9920d..8497866 100644 > --- a/hw/i386/ssdt-proc.hex.generated > +++ b/hw/i386/ssdt-proc.hex.generated > @@ -1,17 +1,26 @@ > +static unsigned char ssdt_proc_end[] = { > +0x8e > +}; > static unsigned char ssdt_proc_name[] = { > 0x28 > }; > +static unsigned char ssdt_proc_pxm[] = { > +0x4e > +}; > +static unsigned char ssdt_proc_id[] = { > +0x38 > +}; > static unsigned char ssdp_proc_aml[] = { > 0x53, > 0x53, > 0x44, > 0x54, > -0x78, > +0x8e, > 0x0, > 0x0, > 0x0, > 0x1, > -0xb8, > +0x19, > 0x42, > 0x58, > 0x50, > @@ -34,21 +43,21 @@ static unsigned char ssdp_proc_aml[] = { > 0x4e, > 0x54, > 0x4c, > -0x23, > -0x8, > -0x13, > +0x28, > +0x5, > +0x10, > 0x20, > 0x5b, > 0x83, > -0x42, > -0x5, > +0x48, > +0x6, > 0x43, > 0x50, > 0x41, > 0x41, > 0xaa, > -0x10, > -0xb0, > +0x0, > +0x0, > 0x0, > 0x0, > 0x0, > @@ -74,6 +83,16 @@ static unsigned char ssdp_proc_aml[] = { > 0x30, > 0x37, > 0x0, > +0x8, > +0x43, > +0x50, > +0x58, > +0x4d, > +0xc, > +0xaa, > +0xaa, > +0xaa, > +0xaa, > 0x14, > 0xf, > 0x5f, > @@ -121,13 +140,19 @@ static unsigned char ssdp_proc_aml[] = { > 0x44, > 0x5f, > 0x5f, > -0x68 > -}; > -static unsigned char ssdt_proc_id[] = { > -0x38 > -}; > -static unsigned char ssdt_proc_end[] = { > -0x78 > +0x68, > +0x14, > +0xb, > +0x5f, > +0x50, > +0x58, > +0x4d, > +0x0, > +0xa4, > +0x43, > +0x50, > +0x58, > +0x4d > }; > static unsigned char ssdt_proc_start[] = { > 0x24 Reviewed-By: Igor Mammedov <imammedo(a)redhat.com>

1 0

Re: [SeaBIOS] [RFC PATCH v2] i386: Add _PXM ACPI method to CPU objects
by Igor Mammedov 29 Nov '13

29 Nov '13

On Wed, 27 Nov 2013 14:02:43 +0100 Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> wrote: > This patch adds a _PXM method to ACPI CPU objects for the pc machine. The _PXM > value is derived from the passed in guest info, same way as CPU SRAT entries. > > Currently, CPU SRAT entries are only enabled for cpus that are already present > in the system. The SRAT entries for hotpluggable processors are disabled (flags > bit 0 set to 0 in hw/i385/acpi-build.c:build_srat). Section 5.2.16.1 of ACPI > spec mentions "If the Local APIC ID of a dynamically added processor is not > present in the SRAT, a _PXM object must exist for the processor’s device or one > of its ancestors in the ACPI Namespace." Since SRAT entries are not available > for the hot-pluggable processors, a _PXM method must exist for them. Otherwise, > the CPU is hot-added in the wrong NUMA node (default node 0). > > Even if CPU SRAT entries are enabled, _PXM method is what the linux kernel > consults on hot-add time. Section 17.2.1 of ACPI spec mentions " OSPM will > consume the SRAT only at boot time. OSPM should use _PXM for any devices that > are hot-added into the system after boot up." To be more precise if SRAT > information is available to the guest kernel, it is used. However, parsed SRAT > info is reset and lost after hot-remove operations, see kernel commit c4c60524. > This means that on a hot-unplug / hot-replug scenario, and without a _PXM > method, the kernel may put a CPU on different nodes because SRAT info has been > reset by a previous hot-remove operation. > > The above hot-remove/hot-add scenario has been tested on master, plus cpu-del > patches from: > https://lists.gnu.org/archive/html/qemu-devel/2013-10/msg01085.html > With the curret _PXM patch, hot-added CPUs are always placed into the correct > NUMA node, regardless of kernel behaviour. > > v1->v2: > Make method return a DWORD integer > Tested on qemu master + cpu-del patches > > Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> > Reviewed-by: Thilo Fromm <t-lo(a)thilo-fromm.de> Patch looks good, Please add patch to update hw/i386/ssdt-proc.hex.generated for hosts without iasl for completness > > --- > hw/i386/acpi-build.c | 5 +++++ > hw/i386/ssdt-proc.dsl | 5 +++++ > 2 files changed, 10 insertions(+) > > diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c > index d089e1e..3c11ddc 100644 > --- a/hw/i386/acpi-build.c > +++ b/hw/i386/acpi-build.c > @@ -603,6 +603,7 @@ static inline char acpi_get_hex(uint32_t val) > #define ACPI_PROC_OFFSET_CPUHEX (*ssdt_proc_name - *ssdt_proc_start + 2) > #define ACPI_PROC_OFFSET_CPUID1 (*ssdt_proc_name - *ssdt_proc_start + 4) > #define ACPI_PROC_OFFSET_CPUID2 (*ssdt_proc_id - *ssdt_proc_start) > +#define ACPI_PROC_OFFSET_CPUPXM (*ssdt_proc_pxm - *ssdt_proc_start) > #define ACPI_PROC_SIZEOF (*ssdt_proc_end - *ssdt_proc_start) > #define ACPI_PROC_AML (ssdp_proc_aml + *ssdt_proc_start) > > @@ -724,6 +725,10 @@ build_ssdt(GArray *table_data, GArray *linker, > proc[ACPI_PROC_OFFSET_CPUHEX+1] = acpi_get_hex(i); > proc[ACPI_PROC_OFFSET_CPUID1] = i; > proc[ACPI_PROC_OFFSET_CPUID2] = i; > + proc[ACPI_PROC_OFFSET_CPUPXM] = guest_info->node_cpu[i]; > + proc[ACPI_PROC_OFFSET_CPUPXM + 1] = 0; > + proc[ACPI_PROC_OFFSET_CPUPXM + 2] = 0; > + proc[ACPI_PROC_OFFSET_CPUPXM + 3] = 0; > } > > /* build this code: > diff --git a/hw/i386/ssdt-proc.dsl b/hw/i386/ssdt-proc.dsl > index 8229bfd..8d4c5bf 100644 > --- a/hw/i386/ssdt-proc.dsl > +++ b/hw/i386/ssdt-proc.dsl > @@ -47,6 +47,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) > * also updating the C code. > */ > Name(_HID, "ACPI0007") > + ACPI_EXTRACT_NAME_DWORD_CONST ssdt_proc_pxm > + Name(PXM, 0xAAAAAAAA) > External(CPMA, MethodObj) > External(CPST, MethodObj) > External(CPEJ, MethodObj) > @@ -59,5 +61,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) > Method(_EJ0, 1, NotSerialized) { > CPEJ(ID, Arg0) > } > + Method(_PXM, 0) { > + Return (PXM) > + } > } > }

3 2

[PATCH v3] i386: Add _PXM ACPI method to CPU objects
by Vasilis Liaskovitis 27 Nov '13

27 Nov '13

This patch adds a _PXM method to ACPI CPU objects for the pc machine. The _PXM value is derived from the passed in guest info, same way as CPU SRAT entries. Currently, CPU SRAT entries are only enabled for cpus that are already present in the system. The SRAT entries for hotpluggable processors are disabled (flags bit 0 set to 0 in hw/i385/acpi-build.c:build_srat). Section 5.2.16.1 of ACPI spec mentions "If the Local APIC ID of a dynamically added processor is not present in the SRAT, a _PXM object must exist for the processor’s device or one of its ancestors in the ACPI Namespace." Since SRAT entries are not available for the hot-pluggable processors, a _PXM method must exist for them. Otherwise, the CPU is hot-added in the wrong NUMA node (default node 0). Even if CPU SRAT entries are enabled, _PXM method is what the linux kernel consults on hot-add time. Section 17.2.1 of ACPI spec mentions " OSPM will consume the SRAT only at boot time. OSPM should use _PXM for any devices that are hot-added into the system after boot up." To be more precise if SRAT information is available to the guest kernel, it is used. However, parsed SRAT info is reset and lost after hot-remove operations, see kernel commit c4c60524. This means that on a hot-unplug / hot-replug scenario, and without a _PXM method, the kernel may put a CPU on different nodes because SRAT info has been reset by a previous hot-remove operation. The above hot-remove/hot-add scenario has been tested on master, plus cpu-del patches from: https://lists.gnu.org/archive/html/qemu-devel/2013-10/msg01085.html With the curret _PXM patch, hot-added CPUs are always placed into the correct NUMA node, regardless of kernel behaviour. v1->v2: Make method return a DWORD integer Tested on qemu master + cpu-del patches v2->v3: Add changed hw/i386/sdt-proc.hex.generated file Change PXM constant name to CPXM Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> Reviewed-by: Thilo Fromm <t-lo(a)thilo-fromm.de> --- hw/i386/acpi-build.c | 5 ++++ hw/i386/ssdt-proc.dsl | 5 ++++ hw/i386/ssdt-proc.hex.generated | 57 ++++++++++++++++++++++++++++----------- 3 files changed, 51 insertions(+), 16 deletions(-) diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c index b48c930..387a869 100644 --- a/hw/i386/acpi-build.c +++ b/hw/i386/acpi-build.c @@ -605,6 +605,7 @@ static inline char acpi_get_hex(uint32_t val) #define ACPI_PROC_OFFSET_CPUHEX (*ssdt_proc_name - *ssdt_proc_start + 2) #define ACPI_PROC_OFFSET_CPUID1 (*ssdt_proc_name - *ssdt_proc_start + 4) #define ACPI_PROC_OFFSET_CPUID2 (*ssdt_proc_id - *ssdt_proc_start) +#define ACPI_PROC_OFFSET_CPUPXM (*ssdt_proc_pxm - *ssdt_proc_start) #define ACPI_PROC_SIZEOF (*ssdt_proc_end - *ssdt_proc_start) #define ACPI_PROC_AML (ssdp_proc_aml + *ssdt_proc_start) @@ -726,6 +727,10 @@ build_ssdt(GArray *table_data, GArray *linker, proc[ACPI_PROC_OFFSET_CPUHEX+1] = acpi_get_hex(i); proc[ACPI_PROC_OFFSET_CPUID1] = i; proc[ACPI_PROC_OFFSET_CPUID2] = i; + proc[ACPI_PROC_OFFSET_CPUPXM] = guest_info->node_cpu[i]; + proc[ACPI_PROC_OFFSET_CPUPXM + 1] = 0; + proc[ACPI_PROC_OFFSET_CPUPXM + 2] = 0; + proc[ACPI_PROC_OFFSET_CPUPXM + 3] = 0; } /* build this code: diff --git a/hw/i386/ssdt-proc.dsl b/hw/i386/ssdt-proc.dsl index 8229bfd..52b44e3 100644 --- a/hw/i386/ssdt-proc.dsl +++ b/hw/i386/ssdt-proc.dsl @@ -47,6 +47,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) * also updating the C code. */ Name(_HID, "ACPI0007") + ACPI_EXTRACT_NAME_DWORD_CONST ssdt_proc_pxm + Name(CPXM, 0xAAAAAAAA) External(CPMA, MethodObj) External(CPST, MethodObj) External(CPEJ, MethodObj) @@ -59,5 +61,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) Method(_EJ0, 1, NotSerialized) { CPEJ(ID, Arg0) } + Method(_PXM, 0) { + Return (CPXM) + } } } diff --git a/hw/i386/ssdt-proc.hex.generated b/hw/i386/ssdt-proc.hex.generated index bb9920d..8497866 100644 --- a/hw/i386/ssdt-proc.hex.generated +++ b/hw/i386/ssdt-proc.hex.generated @@ -1,17 +1,26 @@ +static unsigned char ssdt_proc_end[] = { +0x8e +}; static unsigned char ssdt_proc_name[] = { 0x28 }; +static unsigned char ssdt_proc_pxm[] = { +0x4e +}; +static unsigned char ssdt_proc_id[] = { +0x38 +}; static unsigned char ssdp_proc_aml[] = { 0x53, 0x53, 0x44, 0x54, -0x78, +0x8e, 0x0, 0x0, 0x0, 0x1, -0xb8, +0x19, 0x42, 0x58, 0x50, @@ -34,21 +43,21 @@ static unsigned char ssdp_proc_aml[] = { 0x4e, 0x54, 0x4c, -0x23, -0x8, -0x13, +0x28, +0x5, +0x10, 0x20, 0x5b, 0x83, -0x42, -0x5, +0x48, +0x6, 0x43, 0x50, 0x41, 0x41, 0xaa, -0x10, -0xb0, +0x0, +0x0, 0x0, 0x0, 0x0, @@ -74,6 +83,16 @@ static unsigned char ssdp_proc_aml[] = { 0x30, 0x37, 0x0, +0x8, +0x43, +0x50, +0x58, +0x4d, +0xc, +0xaa, +0xaa, +0xaa, +0xaa, 0x14, 0xf, 0x5f, @@ -121,13 +140,19 @@ static unsigned char ssdp_proc_aml[] = { 0x44, 0x5f, 0x5f, -0x68 -}; -static unsigned char ssdt_proc_id[] = { -0x38 -}; -static unsigned char ssdt_proc_end[] = { -0x78 +0x68, +0x14, +0xb, +0x5f, +0x50, +0x58, +0x4d, +0x0, +0xa4, +0x43, +0x50, +0x58, +0x4d }; static unsigned char ssdt_proc_start[] = { 0x24 -- 1.7.10.4

1 0

[RFC PATCH v2] i386: Add _PXM ACPI method to CPU objects
by Vasilis Liaskovitis 27 Nov '13

27 Nov '13

This patch adds a _PXM method to ACPI CPU objects for the pc machine. The _PXM value is derived from the passed in guest info, same way as CPU SRAT entries. Currently, CPU SRAT entries are only enabled for cpus that are already present in the system. The SRAT entries for hotpluggable processors are disabled (flags bit 0 set to 0 in hw/i385/acpi-build.c:build_srat). Section 5.2.16.1 of ACPI spec mentions "If the Local APIC ID of a dynamically added processor is not present in the SRAT, a _PXM object must exist for the processor’s device or one of its ancestors in the ACPI Namespace." Since SRAT entries are not available for the hot-pluggable processors, a _PXM method must exist for them. Otherwise, the CPU is hot-added in the wrong NUMA node (default node 0). Even if CPU SRAT entries are enabled, _PXM method is what the linux kernel consults on hot-add time. Section 17.2.1 of ACPI spec mentions " OSPM will consume the SRAT only at boot time. OSPM should use _PXM for any devices that are hot-added into the system after boot up." To be more precise if SRAT information is available to the guest kernel, it is used. However, parsed SRAT info is reset and lost after hot-remove operations, see kernel commit c4c60524. This means that on a hot-unplug / hot-replug scenario, and without a _PXM method, the kernel may put a CPU on different nodes because SRAT info has been reset by a previous hot-remove operation. The above hot-remove/hot-add scenario has been tested on master, plus cpu-del patches from: https://lists.gnu.org/archive/html/qemu-devel/2013-10/msg01085.html With the curret _PXM patch, hot-added CPUs are always placed into the correct NUMA node, regardless of kernel behaviour. v1->v2: Make method return a DWORD integer Tested on qemu master + cpu-del patches Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> Reviewed-by: Thilo Fromm <t-lo(a)thilo-fromm.de> --- hw/i386/acpi-build.c | 5 +++++ hw/i386/ssdt-proc.dsl | 5 +++++ 2 files changed, 10 insertions(+) diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c index d089e1e..3c11ddc 100644 --- a/hw/i386/acpi-build.c +++ b/hw/i386/acpi-build.c @@ -603,6 +603,7 @@ static inline char acpi_get_hex(uint32_t val) #define ACPI_PROC_OFFSET_CPUHEX (*ssdt_proc_name - *ssdt_proc_start + 2) #define ACPI_PROC_OFFSET_CPUID1 (*ssdt_proc_name - *ssdt_proc_start + 4) #define ACPI_PROC_OFFSET_CPUID2 (*ssdt_proc_id - *ssdt_proc_start) +#define ACPI_PROC_OFFSET_CPUPXM (*ssdt_proc_pxm - *ssdt_proc_start) #define ACPI_PROC_SIZEOF (*ssdt_proc_end - *ssdt_proc_start) #define ACPI_PROC_AML (ssdp_proc_aml + *ssdt_proc_start) @@ -724,6 +725,10 @@ build_ssdt(GArray *table_data, GArray *linker, proc[ACPI_PROC_OFFSET_CPUHEX+1] = acpi_get_hex(i); proc[ACPI_PROC_OFFSET_CPUID1] = i; proc[ACPI_PROC_OFFSET_CPUID2] = i; + proc[ACPI_PROC_OFFSET_CPUPXM] = guest_info->node_cpu[i]; + proc[ACPI_PROC_OFFSET_CPUPXM + 1] = 0; + proc[ACPI_PROC_OFFSET_CPUPXM + 2] = 0; + proc[ACPI_PROC_OFFSET_CPUPXM + 3] = 0; } /* build this code: diff --git a/hw/i386/ssdt-proc.dsl b/hw/i386/ssdt-proc.dsl index 8229bfd..8d4c5bf 100644 --- a/hw/i386/ssdt-proc.dsl +++ b/hw/i386/ssdt-proc.dsl @@ -47,6 +47,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) * also updating the C code. */ Name(_HID, "ACPI0007") + ACPI_EXTRACT_NAME_DWORD_CONST ssdt_proc_pxm + Name(PXM, 0xAAAAAAAA) External(CPMA, MethodObj) External(CPST, MethodObj) External(CPEJ, MethodObj) @@ -59,5 +61,8 @@ DefinitionBlock ("ssdt-proc.aml", "SSDT", 0x01, "BXPC", "BXSSDT", 0x1) Method(_EJ0, 1, NotSerialized) { CPEJ(ID, Arg0) } + Method(_PXM, 0) { + Return (PXM) + } } } -- 1.7.10.4

1 0

os x boot broken by commit 11948748495841bd54721b250d68c7b3cb0475ef
by Gabriel L. Somlo 26 Nov '13

26 Nov '13

Hi, I'm working on getting OS X working with QEMU/KVM, and just ran into a bit of entropy. I used to be able to boot using Chameleon, and a command line like this: bin/qemu-system-x86_64 -enable-kvm -m 2048 -cpu core2duo \ -usb -device usb-kbd -device usb-mouse \ -bios bios-mac.bin -kernel ./chameleon_2.0_boot -device isa-applesmc,osk="..." \ -M q35 -acpitable file=./seabios/out/src/fw/q35-acpi-dsdt.aml ... I am patching the DSDT source, hence the -acpitable file=... argument, to get QEMU and SeaBIOS to use the modified table. After commit 11948748495841bd54721b250d68c7b3cb0475ef (as of now, the very last one in the git master), this no longer works. I get the xnu boot process hanging with "Waiting for DSMOS", and no progress beyond that. After a cursory look at the commit, I'm wondering if I'm missing some new command line argument to quemu by which I can direct it at the modifed .aml for the DSDT ? It appears to me that I'm unwittingly using the unpatched version... I'm using the latest git master of qemu as well. Any clues you might be able to throw my way much appreciated ! Thanks, --Gabriel

3 7

Re: [SeaBIOS] iasl compiler/disassembler getting confused by Scope operator
by Michael S. Tsirkin 22 Nov '13

22 Nov '13

On Wed, Nov 20, 2013 at 05:25:39PM +0000, Moore, Robert wrote: > If there are still issues, please send the acpidump for the machine. > Maybe you want SSDT.dsl and DSDT.dsl as attachments and not inline?

2 2

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS November 2013