SeaBIOS September 2014

seabios@seabios.org

10 participants
20 discussions

Release ratonale
by Leonardo Passos 14 Sep '14

14 Sep '14

Dear all, I am using Seabios for a personal project and I need to try different versions of it. After cloning the project code, I see many releases when issuing git tag, but which ones are stable? Clearly, 1.7.5-rc1 isn't, but what about the others? Stated otherwise, what is the rationale of release numbering in Seabios? Regards, Leonardo.

1 0

Qemu vgabioses
by Denis 'GNUtoo' Carikli 13 Sep '14

13 Sep '14

Hi, Here is a patch for discussing how to best make qemu vgabioses selectable for all targets, while not making it confusing for users. Denis.

4 11

[PATCH 0/7] Rework USB device detect timing
by Kevin O'Connor 11 Sep '14

11 Sep '14

This series reworks the timing of USB detection. Previously, the USB controllers would wait for the port power to stabilize and then immiediately check for device presence. Unfortunately, the USB specification (USB2 spec) allows for devices to wait up to 100ms from the time that power is stable to the time they assert their presence. This series changes the controllers to behave more like the existing usb-hub code - each port will repeatedly re-check for device presence for 100ms after power is stable. This may also help with USB3 device detection. In particular there were reports of devices that start off identifying as USB2 and then later change to a USB3 type device. The additional polling may catch these cases. This series could cause a slight increase in overall boot time (due to the additional 100ms polling). However, when SeaBIOS' threads are enabled, I think only a few unusual corner cases would cause higher boot times. The series is a fairly significant change to the initial USB timing behaviour. I ran tests on a bunch of USB devices and did not see any regressions. More testing would be appreciated. The series is also at: https://github.com/KevinOConnor/seabios/tree/testing -Kevin Kevin O'Connor (7): ehci: Move port power up from ehci_hub_detect() to check_ehci_ports(). usb-hub: Enable power to all ports prior to calling usb_enumerate(). xhci: Change xhci_hub_detect() to use connect status instead of link state. uhci: Repeatedly poll for device detect for 100ms. ohci: Repeatedly poll for device detect for 100ms. ehci: Stall uhci/ohci init only until default port routing is done. usb: Perform device detect polling on all usb controllers. src/hw/usb-ehci.c | 58 +++++++++++++++++++++++++------------------------------ src/hw/usb-hub.c | 46 +++++++++++++++---------------------------- src/hw/usb-ohci.c | 8 +------- src/hw/usb-uhci.c | 7 +++---- src/hw/usb-xhci.c | 16 +++++---------- src/hw/usb.c | 21 ++++++++++++++------ src/hw/usb.h | 2 +- 7 files changed, 67 insertions(+), 91 deletions(-) -- 1.9.3

2 8

[RFC PATCH 1/1] TPM: Support Physical Presence Interface Spec
by Stefan Berger 11 Sep '14

11 Sep '14

From: Stefan Berger <stefanb(a)linux.vnet.ibm.com> To (try to) avoid SeaBIOS from having to implement a menu for managing the various states of a TPM device, implement the TCG Physical Presence Interface Specification that allows a root user on Linux (for example) to set the opcode for a sequence of TPM operations that the BIOS is supposed to execute upon reboot of the physical or virtual machine. A sequence of operation may for example involve giving up ownership of the TPM and activating and enabling the device. The sequences of operations are defined in table 2 in the specs to be found at the following link: http://www.trustedcomputinggroup.org/resources/tcg_physical_presence_interf… As an example, in recent versions of Linux the opcode (5) can be set as follows: cd /sys/devices/pnp0/00\:04/ppi echo 5 > ppi This ACPI implementation writes the opcode into the TPM's NVRAM at index 0x5001 0000 (to be adapted), which SeaBIOS can then read and react to. An alernative would be to have ACPI write the byte into a memory area that SeaBIOS would have to be able to find upon reboot. More than a single byte of persistent memory will be needed once SeaBIOS supports this since it must store a status byte about the last operation, the opcode that was acted on, and the return code that the TPM gave (see ACPI function 5). Only a subset of the defined opcodes will be supported. Tentatively opcodes 1-11, 14 will be supported. Signed-off-by: Stefan Berger <stefanb(a)linux.vnet.ibm.com> --- hw/i386/ssdt-tpm.dsl | 208 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 208 insertions(+) diff --git a/hw/i386/ssdt-tpm.dsl b/hw/i386/ssdt-tpm.dsl index 75d9691..678b212 100644 --- a/hw/i386/ssdt-tpm.dsl +++ b/hw/i386/ssdt-tpm.dsl @@ -12,6 +12,14 @@ * You should have received a copy of the GNU General Public License along * with this program; if not, see <http://www.gnu.org/licenses/>. */ + +/* + * Note: This implementation follows the TCG specification + * 'TCG Physical Presence Interface Specification' of which the latest + * version can be found here: + * http://www.trustedcomputinggroup.org/resources/tcg_physical_presence_interf… + */ + #include "hw/acpi/tpm.h" ACPI_EXTRACT_ALL_CODE ssdt_tpm_aml @@ -38,6 +46,206 @@ DefinitionBlock ( Method (_STA, 0, NotSerialized) { Return (0x0F) } + // Last accepted opcode + NAME(OP, Zero) + OperationRegion (TTIS, SystemMemory, + TPM_TIS_ADDR_BASE, TPM_TIS_ADDR_SIZE) + + // Define TIS registers + Field(TTIS, AnyAcc, NoLock, Preserve) { + Offset (0x00), + TACC, 8, // Access register + Offset (0x18), + TSTS, 8, // Status register + Offset (0x24), + TDAT, 8, // Data register + Offset (0xf90), + TDBG, 32 // QEMU TIS Debug + } + // TPM_NV_WriteValue for writing 1 byte to a NVRAM index + Name (NVWR, Package() { + 0x00, 0xc1, + 0x00, 0x00, 0x00, 0x17, // total length + 0x00, 0x00, 0x00, 0xcd, // opcode + 0x50, 0x01, 0x00, 0x00, // index + 0x00, 0x00, 0x00, 0x00, // offset + 0x00, 0x00, 0x00, 0x01, // payload length + 0x00 + }) + // Buffer to read TPM responses into + Name (BUFF, Buffer(0xa){}) + // Last TPM return code from writing to TPM NVRAM + Name (WRET, 0) + + // Write into NVRAM of TPM + Method ( _WNV, 1, NotSerialized) { + // Any early return will indicate a failed execution + Store(0xffffffff, WRET) + + // Request access to locty 0 + Store(0x02, TACC) + + // Make sure we are active + Store(TACC, Local0) + If (LNotEqual(And(Local0, 0xa0), 0xa0)) { + return (16) // EBUSY + } + + // Set the Command Ready Flag + Store(0x40, TSTS) + + // Make sure we are ready + Store(TSTS, Local0) + If (LNotEqual(And(Local0, 0x40), 0x40)) { + return (14) // EFAULT + } + + // Write the user-requested op-code into the NVRAM command + Store(Arg0, Index(NVWR, 0x16)) + + // Write the TPM_NV_WriteValue into the data buffer + Store(0x0, Local0) + While (LLess(Local0, SizeOf(NVWR))) { + Store(DerefOf(Index(NVWR, Local0)), TDAT) + Increment(Local0) + } + + // Go: have TPM process the command + Store(0x20, TSTS) + + // Wait for response to be ready + Store(0, Local0) + While (LLess(local0, 0x30)) { + Store(TSTS, Local1) + if (LEqual(And(Local1, 0x10), 0x10)) { + break + } + Sleep(10) + Increment(Local0) + } + + // Read the result into BUFF + Store(0, Local0) + While (LLess(Local0, SizeOf(BUFF))) { + Store(TDAT, Index(BUFF, Local0)) + Increment(Local0) + } + + // Extract return code from response + Mid(BUFF, 6, 4, WRET) + + Return (0) + } + Method (_DSM, 4, NotSerialized) { + If (LEqual (Arg0, ToUUID("3DDDFAA6-361B-4EB4-A424-8D10089D1653"))) { + + // only supporting API revision 1 + If (LNotEqual (Arg1, 1)) { + Return (Buffer (1) {0}) + } + + Store(ToInteger(Arg2), Local0) + // standard DSM query function + If (LEqual (Local0, 0)) { + Return (Buffer () {0xFF, 0x01}) + } + + // interface version + If (LEqual (Local0, 1)) { + Return ("1.2") + } + + // submit TPM operation + If (LEqual (Local0, 2)) { + // get opcode from package + Store(DerefOf(Index(Arg3, 0)), Local0) + // check for supported opcode + // supported opcodes: 1-11, 14 + If (LOr(LAnd(LGreaterEqual(Local0, 1), LLessEqual(Local0, 11)), LEqual(Local0, 14))) { + // Write the OP into TPM NVRAM + Store(_WNV ( Local0 ), Local1) + if (LNotEqual(Local1, 0)) { + return (Local1) + } + // Check for error during NVRAM write + if (LEqual(WRET, 0)) { + Store(Local0, OP) + Return (0) + } else { + Return (1) + } + } else { + Return (1) + } + } + + // get pending TPM operation + If (LEqual (Local0, 3)) { + NAME(PEOP, Package(2) { 0, 0 }) + + // Check for error during NVRAM write + if (LEqual(WRET, 0)) { + Store ( 0, Index(PEOP, 0)) + } else { + Store ( 1, Index(PEOP, 0)) + } + Store ( OP, Index(PEOP, 1)) + + Return (PEOP) + } + + // action to transition to pre-OS env. + If (LEqual (Local0, 4)) { + return (2) // Requiring reboot + } + + // get pre-OS TPM operation response + If (LEqual (Local0, 5)) { + // missing BIOS support + Name (OPRE, Package(3) { 1, 0, 0}) + return (OPRE) + } + + // preferred user language + If (LEqual (Local0, 6)) { + return (3) // Not implemented + } + + // submit TPM operation v2 + If (LEqual (Local0, 7)) { + Store(DerefOf(Index(Arg3, 0)), Local0) + // supported opcodes: 1-11, 14 + If (LOr(LAnd(LGreaterEqual(Local0, 1), LLessEqual(Local0, 11)), LEqual(Local0, 14))) { + // Write the OP into TPM NVRAM + Store(_WNV ( Local0 ), Local1) + if (LNotEqual(Local1, 0)) { + return (Local1) + } + // Check for error during NVRAM write + if (LEqual(WRET, 0)) { + // Remember opcode locally + Store(Local0, OP) + Return (0) + } else { + Return (1) + } + } else { + Return (1) + } + } + + // get user confirmation status + If (LEqual (Local0, 8)) { + Store(DerefOf(Index(Arg3,0)), Local0) + If (LLessEqual(Local0, 22)) { + Return (4) // allowed, no user required + } else { + Return (0) // not implemented + } + } + } + return (Buffer() { 0x0 }) + } } } } -- 1.9.3

2 1

[PATCH 0/4] xhci code cleanups and some reorganization
by Kevin O'Connor 10 Sep '14

10 Sep '14

This series cleans up the xhci code (by removing some of the old 16bit assumptions), and rearranges things a little to make the code layout more similar to the other usb controller drivers. This is in preparation for the next USB series which reworks some of the USB timing. This series (along with the next) is also at: https://github.com/KevinOConnor/seabios/tree/testing -Kevin Kevin O'Connor (4): xhci: Remove 16bit code wrappers. xhci: Use high memory instead of low memory for internal storage. xhci: Move root hub and setup code to top of file. xhci: Add xhci_check_ports() and xhci_free_pipes() functions. src/hw/usb-xhci.c | 924 +++++++++++++++++++++++++++--------------------------- 1 file changed, 462 insertions(+), 462 deletions(-) -- 1.9.3

1 4

Current tip not booting from USB or SD cards
by Stefan Berger 10 Sep '14

10 Sep '14

Hello! The current git tip does not boot from USB or SD cards on an Acer C720. The issue seems to stem from: ab9d771ce14d5b8d38cde373d15a51b45efb99c6 is the first bad commit commit ab9d771ce14d5b8d38cde373d15a51b45efb99c6 Author: Kevin O'Connor <kevin(a)koconnor.net> Date: Sat Jun 14 12:48:35 2014 -0400 ehci: Update usb command timeouts to use usb_xfer_time() Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> :040000 040000 4fcdb230f8fcffc0d8256225eb3c3220bd2293d4 78b62cb78cc3537da5a2ce1f1417fb1e0a571ba1 M src Regards, Stefan

2 2

[PATCH 1/2] usb: Fix usb_xfer_time() to work when called in 16bit mode.
by Kevin O'Connor 10 Sep '14

10 Sep '14

Make sure to wrap accesses to the usb_pipe struct with GET_LOWFLAT so that it works in 16bit mode. This bug impacts both ehci and uhci usb controllers (it should not impact ohci and xhci as those never call the function in 16bit mode). Signed-off-by: Kevin O'Connor <kevin(a)koconnor.net> --- src/hw/usb.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/hw/usb.c b/src/hw/usb.c index 46b46be..930b5d9 100644 --- a/src/hw/usb.c +++ b/src/hw/usb.c @@ -195,7 +195,7 @@ usb_xfer_time(struct usb_pipe *pipe, int datalen) // set_address commands where we don't want to stall the boot if // the device doesn't actually exist. Add 100ms to account for // any controller delays. - if (!pipe->devaddr) + if (!GET_LOWFLAT(pipe->devaddr)) return USB_TIME_STATUS + 100; return USB_TIME_COMMAND + 100; } -- 1.9.3

2 2

Re: [SeaBIOS] EHCI boot problems
by Kevin O'Connor 10 Sep '14

10 Sep '14

On Tue, Sep 09, 2014 at 04:08:25PM -0600, Dave Frodin wrote: > Kevin, > With your patch I'm not seeing that timeout message get displayed. > I also forgot to mention that the USB thumbdrive will always get identified > by SeaBIOS and will be listed as a boot device but will hang as soon as it > is selected in the F12 boot menu. The activity LED on the thumbdrive only > flashes when Seabios is identifying it. I was able to reproduce the problem on my c720. The patch below seems to fix it for me (but I'm still running tests). I'm not sure why this problem didn't show up on my other test machine (e350m1). -Kevin --- a/src/hw/usb.c +++ b/src/hw/usb.c @@ -195,7 +195,7 @@ usb_xfer_time(struct usb_pipe *pipe, int datalen) // set_address commands where we don't want to stall the boot if // the device doesn't actually exist. Add 100ms to account for // any controller delays. - if (!pipe->devaddr) + if (!GET_LOWFLAT(pipe->devaddr)) return USB_TIME_STATUS + 100; return USB_TIME_COMMAND + 100; }

2 1

Re: [SeaBIOS] EHCI boot problems
by Kevin O'Connor 10 Sep '14

10 Sep '14

On Tue, Sep 09, 2014 at 01:15:04PM -0600, Dave Frodin wrote: > I'm seeing a problem booting from USB thumbdrives with commit ab9d771ce > ehci: Update usb command timeouts to use usb_xfer_time() > > I'm not quite sure what the problem is other than it not liking the new > timeouts. > I couldn't see any problems with the ehci_control() calls to ehci_wait_td(). > But the ehci_send_bulk() only allows the system to boot if I change > int ret = ehci_wait_td(pipe, td, end); > to > int ret = ehci_wait_td(pipe, td, timer_calc(5000)); > > One potential problem is that the "end" value is calculated once and reused > multiple times in the functions. Prior to the commit the timeout value was > passed > to the ehci_wait_td() function. Now the final "end" time is passed. So it > looks like > once the end timeout is reached in one of the loops that calls > ehci_wait_td() the > timer will always be expired for any other calls from that function. > > One solution might be to get rid of the "end" variable and change the calls > from > ret = ehci_wait_td(pipe, td, end); > to > ret = ehci_wait_td(pipe, td, timer_calc(usb_xfer_time(p, datasize))); > > Making the above change didn't have any impact on my booting problem until > I forced the ehci_send_bulk() timeouts to 5000. Hi Dave, Stefan Burger reported a similar problem with that commit. However, I'm puzzled what would be causing such a regression - five seconds should be more then enough time to complete a single usb transaction. Do you have serial debugging (or similar) available on the device you are working with? If you could provide the output from the debugging patch below it may help diagnose the issue. I also got my Acer c720 back, and will run some tests on it. -Kevin --- a/src/hw/usb-ehci.c +++ b/src/hw/usb-ehci.c @@ -629,6 +629,7 @@ ehci_send_bulk(struct usb_pipe *p, int dir, void *data, int datasize) , &pipe->qh, dir, data, datasize); // Allocate 4 tds on stack (with required alignment) + u32 starttime = timer_calc(0); u32 end = timer_calc(usb_xfer_time(p, datasize)); u8 tdsbuf[sizeof(struct ehci_qtd) * STACKQTDS + EHCI_QTD_ALIGN - 1]; struct ehci_qtd *tds = (void*)ALIGN((u32)tdsbuf, EHCI_QTD_ALIGN); @@ -642,7 +643,7 @@ ehci_send_bulk(struct usb_pipe *p, int dir, void *data, int datasize) struct ehci_qtd *td = &tds[tdpos++ % STACKQTDS]; int ret = ehci_wait_td(pipe, td, end); if (ret) - return -1; + goto fail; struct ehci_qtd *nexttd_fl = MAKE_FLATPTR(GET_SEG(SS) , &tds[tdpos % STACKQTDS]); @@ -662,10 +663,15 @@ ehci_send_bulk(struct usb_pipe *p, int dir, void *data, int datasize) struct ehci_qtd *td = &tds[tdpos++ % STACKQTDS]; int ret = ehci_wait_td(pipe, td, end); if (ret) - return -1; + goto fail; } return 0; +fail: + dprintf(1, "timeout: start=%x end=%x cur=%x time=%d devaddr=%x datasize=%d\n" + , starttime, end, timer_calc(0) + , usb_xfer_time(p, datasize), p->devaddr, datasize); + return -1; } int

2 1

EHCI boot problems
by Dave Frodin 09 Sep '14

09 Sep '14

I'm seeing a problem booting from USB thumbdrives with commit ab9d771ce ehci: Update usb command timeouts to use usb_xfer_time() I'm not quite sure what the problem is other than it not liking the new timeouts. I couldn't see any problems with the ehci_control() calls to ehci_wait_td(). But the ehci_send_bulk() only allows the system to boot if I change int ret = ehci_wait_td(pipe, td, end); to int ret = ehci_wait_td(pipe, td, timer_calc(5000)); One potential problem is that the "end" value is calculated once and reused multiple times in the functions. Prior to the commit the timeout value was passed to the ehci_wait_td() function. Now the final "end" time is passed. So it looks like once the end timeout is reached in one of the loops that calls ehci_wait_td() the timer will always be expired for any other calls from that function. One solution might be to get rid of the "end" variable and change the calls from ret = ehci_wait_td(pipe, td, end); to ret = ehci_wait_td(pipe, td, timer_calc(usb_xfer_time(p, datasize))); Making the above change didn't have any impact on my booting problem until I forced the ehci_send_bulk() timeouts to 5000. Thanks in advance for any advice anyone can offer, Dave

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

SeaBIOS September 2014