Skip to contentSkip to conversation

x86: Add support for booting from TPL

PATCH Superseded Simon Glass 4 participants 113 messages Apr 26, 2019 – May 9, 2019

Subsystems (5)
x86 Kbuild PCI Char and misc drivers +1 more subsystems

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v2 cover x86: Add support for booting from TPL

At present SPL is used on 64-bit platforms, to allow SPL to be built as a 32-bit program and U-Boot proper to be built as 64-bit.

However it is useful to be able to use SPL on any x86 platform, where U-Boot needs to be updated in the field. Then SPL can select which U-Boot to run (A or B) and most of the code can be updated. Similarly, using TPL allows both SPL and U-Boot to be updated. This is the best approach, since it means that all of U-Boot proper as well as SPL (in particular SDRAM init) can be updated in the field. This provides for the smallest possible amount of read-only (non-updateable) code: just the TPL code.

This series contains a number of changes to allow x86 boards to use TPL, SPL and U-Boot proper. As a test, it is enabled for samus with a new chromebook_samus_tpl board.

Changes in v2:
- Update commit message to mention dropping the \n
- Update testSelectImage() to test in normal and verbose modes
- Update the commit message to explain the implications on aliases
- Add a better explanation of the logic change, in the commit message
- Fix map output when section offset is not set (make it 0)
- Add a test for sections with offsets
- Update the comment in fsp_cap.S too
- Update commit message to make it clear this patch is just for broadwell
- Bring in sdram_console_tx_byte() to allow debugging
- Add xorl to TPL code also
- Update comments in start_from_tpl to correctly explain SPL state
- Add new patch to separate out the EFI code in sysreset
- Add new patch to add an ioctl to read power-management info
- Add new patch to implement PCH_REQ_PMBASE_INFO on ivybridge
- Add new patch to implement PCH_REQ_PMBASE_INFO on broadwell
- Add new patch to implement power-off if available
- Add new patch to enable the RTC in Kconfig
- Add a new patch to update PCH to work in TPL
- Add a new patch allowing jumping from TPL to SPL
- Sort defconfig and adjust it to build after rebase on maste

Simon Glass (50):
binman: Don't generate an error in 'text' entry constructor
binman: Don't show image-skip message by default
binman: Add a missing comment in Entry_vblock
dm: core: Fix translate condition in ofnode_get_addr_size()
cros_ec: Use a hyphen in the uclass name
spl: Allow sandbox to build a device-tree file
binman: Allow sections to have an offset
x86: start64: Fix copyright message
x86: mp_init: Use proper error numbers
x86: Add a way to reinit the cpu
x86: dts: Add device-tree labels for rtc and reset
x86: Update a stale comment about ifdtool
x86: Support SPL and TPL
x86: Support booting with TPL
x86: Add a handoff header file
x86: broadwell: Improve SDRAM debugging output
x86: broadwell: Allow SDRAM init from SPL
x86: Move init of debug UART to cpu.c
x86: broadwell: Split CPU init
x86: Add support for starting from SPL/TPL
x86: Allow 16-bit init to be in TPL
x86: broadwell: Allow booting from SPL
x86: broadwell: Select refcode and CPU code for SPL
x86: Add common Intel code for SPL
x86: Support saving MRC data from SPL
x86: Add a simple TPL implementation
x86: mrccache: Add more debugging
x86: sysreset: Separate out the EFI code
x86: pch: Add an ioctl to read power-management info
x86: ivybridge: Implement PCH_REQ_PMBASE_INFO
x86: broadwell: Implement PCH_REQ_PMBASE_INFO
x86: sysreset: Implement power-off if available
x86: Support TPL in Intel common code
x86: Don't set up MTRRs in SPL
x86: Don't generate a bootstage report in SPL
x86: Support PCI VGA ROM when TPL is used
x86: sysreset: Implement the get_last() method
x86: Add documention on the samus flashmap
x86: samus: Update device tree for SPL
x86: samus: Update device tree for verified boot
x86: Update device tree for TPL
x86: Update device tree for Chromium OS verified boot
x86: Fix device-tree indentation
x86: samus: Increase the pre-reloc memory again
Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"
x86: Enable the RTC on all boards
x86: Update the memory map a little
x86: broadwell: Update PCH to work in TPL
x86: Add a way to jump from TPL to SPL
x86: samus: Add a target to boot through TPL

 Makefile                                  |   1 +
 arch/Kconfig                              |  32 +
 arch/x86/Kconfig                          |  10 +-
 arch/x86/Makefile                         |  16 +-
 arch/x86/cpu/Makefile                     |  15 +-
 arch/x86/cpu/broadwell/Makefile           |  23 +-
 arch/x86/cpu/broadwell/cpu.c              | 676 +--------------------
 arch/x86/cpu/broadwell/cpu_from_spl.c     |  63 ++
 arch/x86/cpu/broadwell/cpu_full.c         | 694 ++++++++++++++++++++++
 arch/x86/cpu/broadwell/northbridge.c      | 100 ++++
 arch/x86/cpu/broadwell/pch.c              |  37 +-
 arch/x86/cpu/broadwell/sdram.c            | 136 +----
 arch/x86/cpu/i386/cpu.c                   | 113 ++--
 arch/x86/cpu/intel_common/Makefile        |  17 +-
 arch/x86/cpu/intel_common/car.S           |   2 +-
 arch/x86/cpu/intel_common/cpu_from_spl.c  |  27 +
 arch/x86/cpu/ivybridge/bd82x6x.c          |  15 +
 arch/x86/cpu/mp_init.c                    |  10 +-
 arch/x86/cpu/start.S                      |  13 +
 arch/x86/cpu/start64.S                    |   2 +-
 arch/x86/cpu/start_from_spl.S             |  71 +++
 arch/x86/cpu/start_from_tpl.S             |  49 ++
 arch/x86/cpu/u-boot-spl.lds               |   2 +-
 arch/x86/cpu/x86_64/cpu.c                 |   5 +
 arch/x86/dts/chromebook_samus.dts         |  60 +-
 arch/x86/dts/reset.dtsi                   |   2 +-
 arch/x86/dts/rtc.dtsi                     |   2 +-
 arch/x86/dts/u-boot.dtsi                  | 154 +++--
 arch/x86/include/asm/handoff.h            |  15 +
 arch/x86/include/asm/mrccache.h           |  11 +
 arch/x86/include/asm/spl.h                |  17 +-
 arch/x86/include/asm/u-boot-x86.h         |  20 +
 arch/x86/lib/Makefile                     |   9 +-
 arch/x86/lib/bootm.c                      |   2 +-
 arch/x86/lib/fsp/fsp_car.S                |   2 +-
 arch/x86/lib/init_helpers.c               |   5 +-
 arch/x86/lib/mrccache.c                   |  52 +-
 arch/x86/lib/spl.c                        |  44 +-
 arch/x86/lib/tpl.c                        | 118 ++++
 board/google/Kconfig                      |   8 +
 board/google/chromebook_samus/Kconfig     |  14 +-
 board/google/chromebook_samus/MAINTAINERS |   7 +
 configs/chromebook_link_defconfig         |   1 +
 configs/chromebook_samus_defconfig        |   2 +-
 configs/chromebook_samus_tpl_defconfig    |  81 +++
 doc/README.x86                            |  16 +
 drivers/core/ofnode.c                     |   2 +-
 drivers/misc/cros_ec.c                    |   2 +-
 drivers/pci/pci_rom.c                     |   2 +-
 drivers/sysreset/sysreset_x86.c           | 104 +++-
 include/configs/chromebook_link.h         |   3 -
 include/configs/chromebook_samus.h        |   3 +
 include/configs/qemu-x86.h                |   3 -
 include/configs/x86-common.h              |   1 -
 include/pch.h                             |  18 +
 include/pci.h                             |   6 +-
 scripts/Makefile.spl                      |  24 +-
 tools/binman/README                       |   7 +
 tools/binman/bsection.py                  |   9 +-
 tools/binman/control.py                   |   4 +-
 tools/binman/etype/section.py             |   3 +-
 tools/binman/etype/text.py                |   4 +-
 tools/binman/etype/vblock.py              |   1 +
 tools/binman/ftest.py                     |  44 +-
 tools/binman/test/101_sections_offset.dts |  35 ++
 65 files changed, 2066 insertions(+), 980 deletions(-)
 create mode 100644 arch/x86/cpu/broadwell/cpu_from_spl.c
 create mode 100644 arch/x86/cpu/broadwell/cpu_full.c
 create mode 100644 arch/x86/cpu/intel_common/cpu_from_spl.c
 create mode 100644 arch/x86/cpu/start_from_spl.S
 create mode 100644 arch/x86/cpu/start_from_tpl.S
 create mode 100644 arch/x86/include/asm/handoff.h
 create mode 100644 arch/x86/lib/tpl.c
 create mode 100644 configs/chromebook_samus_tpl_defconfig
 create mode 100644 tools/binman/test/101_sections_offset.dts
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#1 lore raw

PATCH v2 1/50 binman: Don't generate an error in 'text' entry constructor

It is not good practice to raise an exception in a constructor. In this case the 'text' entry may not actually be used, if -i is used to filter out the images that get built.

Move the exception to where the data is actually used.

---

Changes in v2: None

 tools/binman/etype/text.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/binman/etype/text.py +2 -2
index 6e99819487f..c4aa510a87b 100644
@@ -51,10 +51,10 @@ class Entry_text(Entry):
         self.text_label, = self.GetEntryArgsOrProps(
             [EntryArg('text-label', str)])
         self.value, = self.GetEntryArgsOrProps([EntryArg(self.text_label, str)])
+
+    def ObtainContents(self):
         if not self.value:
             self.Raise("No value provided for text label '%s'" %
                        self.text_label)
-
-    def ObtainContents(self):
         self.SetContents(self.value)
         return True
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#2 lore raw

PATCH v2 1/50 binman: Don't generate an error in 'text' entry constructor

Quoted 7 lines · Simon Glass

It is not good practice to raise an exception in a constructor. In this case the 'text' entry may not actually be used, if -i is used to filter out the images that get built.

Move the exception to where the data is actually used.

Quoted 9 lines

---

Changes in v2: None

 tools/binman/etype/text.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

applied to u-boot-x86, thanks!

#3 lore raw

PATCH v2 2/50 binman: Don't show image-skip message by default

In reply to Simon Glass

This message is not very important since it is simply indicating that the user's instructions are being followed. Only show it when the verbosity level is above the default.

Also drop the unnecessary extra newline on this message, which causes two line breaks.

---

Changes in v2:
- Update commit message to mention dropping the \n
- Update testSelectImage() to test in normal and verbose modes

 tools/binman/control.py |  4 ++--
 tools/binman/ftest.py   | 26 +++++++++++++++++++-------
 2 files changed, 21 insertions(+), 9 deletions(-)
tools/binman/control.py +2 -2
index 3446e2e79c5..b32e4e1996f 100644
@@ -133,8 +133,8 @@ def Binman(options, args):
                     if name not in options.image:
                         del images[name]
                         skip.append(name)
-                if skip:
-                    print 'Skipping images: %s\n' % ', '.join(skip)
+                if skip and options.verbosity >= 2:
+                    print 'Skipping images: %s' % ', '.join(skip)
 
             state.Prepare(images, dtb)
 
tools/binman/ftest.py +19 -7
index e77fce5a26f..75658c9a366 100644
@@ -187,7 +187,8 @@ class TestFunctional(unittest.TestCase):
         return control.Binman(options, args)
 
     def _DoTestFile(self, fname, debug=False, map=False, update_dtb=False,
-                    entry_args=None, images=None, use_real_dtb=False):
+                    entry_args=None, images=None, use_real_dtb=False,
+                    verbosity=None):
         """Run binman with a given test file
 
         Args:
@@ -210,6 +211,8 @@ class TestFunctional(unittest.TestCase):
             args.append('-up')
         if not use_real_dtb:
             args.append('--fake-dtb')
+        if verbosity is not None:
+            args.append('-v%d' % verbosity)
         if entry_args:
             for arg, value in entry_args.iteritems():
                 args.append('-a%s=%s' % (arg, value))
@@ -1459,13 +1462,22 @@ class TestFunctional(unittest.TestCase):
 
     def testSelectImage(self):
         """Test that we can select which images to build"""
-        with test_util.capture_sys_output() as (stdout, stderr):
-            retcode = self._DoTestFile('006_dual_image.dts', images=['image2'])
-        self.assertEqual(0, retcode)
-        self.assertIn('Skipping images: image1', stdout.getvalue())
+        expected = 'Skipping images: image1'
+
+        # We should only get the expected message in verbose mode
+        for verbosity in (None, 2):
+            with test_util.capture_sys_output() as (stdout, stderr):
+                retcode = self._DoTestFile('006_dual_image.dts',
+                                           verbosity=verbosity,
+                                           images=['image2'])
+            self.assertEqual(0, retcode)
+            if verbosity:
+                self.assertIn(expected, stdout.getvalue())
+            else:
+                self.assertNotIn(expected, stdout.getvalue())
 
-        self.assertFalse(os.path.exists(tools.GetOutputFilename('image1.bin')))
-        self.assertTrue(os.path.exists(tools.GetOutputFilename('image2.bin')))
+            self.assertFalse(os.path.exists(tools.GetOutputFilename('image1.bin')))
+            self.assertTrue(os.path.exists(tools.GetOutputFilename('image2.bin')))
 
     def testUpdateFdtAll(self):
         """Test that all device trees are updated with offset/size info"""
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#4 lore raw

PATCH v2 2/50 binman: Don't show image-skip message by default

Quoted 11 lines · Simon Glass

This message is not very important since it is simply indicating that the user's instructions are being followed. Only show it when the verbosity level is above the default.

Also drop the unnecessary extra newline on this message, which causes two line breaks.

---

Quoted 8 lines

Changes in v2:
- Update commit message to mention dropping the \n
- Update testSelectImage() to test in normal and verbose modes

 tools/binman/control.py |  4 ++--
 tools/binman/ftest.py   | 26 +++++++++++++++++++-------
 2 files changed, 21 insertions(+), 9 deletions(-)
#5 lore raw

PATCH v2 2/50 binman: Don't show image-skip message by default

Quoted 17 lines · Bin Meng
Quoted 15 lines · Simon Glass

This message is not very important since it is simply indicating that the user's instructions are being followed. Only show it when the verbosity level is above the default.

Also drop the unnecessary extra newline on this message, which causes two line breaks.

---

Changes in v2:
- Update commit message to mention dropping the \n
- Update testSelectImage() to test in normal and verbose modes

Quoted 6 lines
Quoted 4 lines
 tools/binman/control.py |  4 ++--
 tools/binman/ftest.py   | 26 +++++++++++++++++++-------
 2 files changed, 21 insertions(+), 9 deletions(-)

applied to u-boot-x86, thanks!

#6 lore raw

PATCH v2 3/50 binman: Add a missing comment in Entry_vblock

In reply to Simon Glass

An important property is missing. Update the entry comment to include this.

---

Changes in v2: None

 tools/binman/etype/vblock.py | 1 +
 1 file changed, 1 insertion(+)
tools/binman/etype/vblock.py +1 -0
index c4d970ed160..334ff9f966a 100644
@@ -18,6 +18,7 @@ class Entry_vblock(Entry):
     """An entry which contains a Chromium OS verified boot block
 
     Properties / Entry arguments:
+        - content: List of phandles to entries to sign
         - keydir: Directory containing the public keys to use
         - keyblock: Name of the key file to use (inside keydir)
         - signprivate: Name of provide key file to use (inside keydir)
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#7 lore raw

PATCH v2 3/50 binman: Add a missing comment in Entry_vblock

Quoted 4 lines · Simon Glass

An important property is missing. Update the entry comment to include this.

Quoted 9 lines

---

Changes in v2: None

 tools/binman/etype/vblock.py | 1 +
 1 file changed, 1 insertion(+)

applied to u-boot-x86, thanks!

#8 lore raw

PATCH v2 4/50 dm: core: Fix translate condition in ofnode_get_addr_size()

In reply to Simon Glass

Update the condition to translate only if this is enabled for SPL.

---

Changes in v2: None

 drivers/core/ofnode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/core/ofnode.c +1 -1
index 785f5c3acf7..cc0c031e0d7 100644
@@ -546,7 +546,7 @@ fdt_addr_t ofnode_get_addr_size(ofnode node, const char *property,
 		ns = of_n_size_cells(np);
 		*sizep = of_read_number(prop + na, ns);
 
-		if (IS_ENABLED(CONFIG_OF_TRANSLATE) && ns > 0)
+		if (CONFIG_IS_ENABLED(OF_TRANSLATE) && ns > 0)
 			return of_translate_address(np, prop);
 		else
 			return of_read_number(prop, na);
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#9 lore raw

PATCH v2 4/50 dm: core: Fix translate condition in ofnode_get_addr_size()

Quoted 12 lines · Simon Glass

Update the condition to translate only if this is enabled for SPL.

---

Changes in v2: None

 drivers/core/ofnode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

applied to u-boot-x86, thanks!

#10 lore raw

PATCH v2 5/50 cros_ec: Use a hyphen in the uclass name

In reply to Simon Glass

Device-tree rules require that aliases use a hyphen rather than a underscore. Update the uclass name to fit with this.

This allows device-tree aliases to be used to refer to cros-ec devices, for example:

	aliases {
	    cros-ec0 = &ec;
	    cros-ec1 = &pd;
	};

---

Changes in v2:
- Update the commit message to explain the implications on aliases

 drivers/misc/cros_ec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/misc/cros_ec.c +1 -1
index 565de040fe9..382f8262863 100644
@@ -1482,7 +1482,7 @@ int cros_ec_set_lid_shutdown_mask(struct udevice *dev, int enable)
 
 UCLASS_DRIVER(cros_ec) = {
 	.id		= UCLASS_CROS_EC,
-	.name		= "cros_ec",
+	.name		= "cros-ec",
 	.per_device_auto_alloc_size = sizeof(struct cros_ec_dev),
 	.post_bind	= dm_scan_fdt_dev,
 	.flags		= DM_UC_FLAG_ALLOC_PRIV_DMA,
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#11 lore raw

PATCH v2 5/50 cros_ec: Use a hyphen in the uclass name

Hi Simon,

Quoted 11 lines · Simon Glass

Device-tree rules require that aliases use a hyphen rather than a underscore. Update the uclass name to fit with this.

This allows device-tree aliases to be used to refer to cros-ec devices, for example:

        aliases {
            cros-ec0 = &ec;
            cros-ec1 = &pd;
        };

Thanks for adding the explanation here. But I was wondering since this is required by DT rules for the aliases node, do we have to require all UCLASS_DRIVERs to have name with hyphen instead of underscore?

Quoted 10 lines

---

Changes in v2:
- Update the commit message to explain the implications on aliases

 drivers/misc/cros_ec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Regards,
Bin

#12 lore raw

PATCH v2 5/50 cros_ec: Use a hyphen in the uclass name

Hi Bin,

Quoted 11 lines · Bin Meng

Hi Simon,

Quoted 7 lines · Simon Glass

Device-tree rules require that aliases use a hyphen rather than a underscore. Update the uclass name to fit with this.

This allows device-tree aliases to be used to refer to cros-ec devices, for example:

Quoted 8 lines
Quoted 4 lines
        aliases {
            cros-ec0 = &ec;
            cros-ec1 = &pd;
        };

Thanks for adding the explanation here. But I was wondering since this is required by DT rules for the aliases node, do we have to require all UCLASS_DRIVERs to have name with hyphen instead of underscore?

It might be good to have this rule, in case people want to use aliases.

Regards,
Simon

#13 lore raw

PATCH v2 6/50 spl: Allow sandbox to build a device-tree file

In reply to Simon Glass

At present only OF_SEPARATE is considered valid for building a device-tree file in SPL. However sandbox uses OF_HOSTFILE instead. Update the logic to handle this and make it easier to understand.

Note that the new logic is not quite the same as the old logic. It was previously assumed that checking for:

   $(CONFIG_$(SPL_TPL_)OF_CONTROL)
   $(CONFIG_OF_SEPARATE)
   $(CONFIG_$(SPL_TPL_)OF_PLATDATA)

producing 'yy' meant that the first two were 'y' and the last was empty. Strictly speaking it would be possible for any two of the three to be 'y' and still yield the same result. However, that was not the intention of the new logic, since OF_PLATDATA always ensures that no device-tree file is included. So in effect the new logic is the same, with the addition of OF_HOSTFILE as an option for OP_SEPARATE.

---

Changes in v2:
- Add a better explanation of the logic change, in the commit message

 scripts/Makefile.spl | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
scripts/Makefile.spl +13 -1
index 56dbbe127ba..8d76a2d4007 100644
@@ -253,8 +253,20 @@ else
 FINAL_DTB_CONTAINER = $(obj)/$(SPL_BIN).multidtb.fit
 endif
 
+# Build the .dtb file if:
+#   - we are not using OF_PLATDATA
+#   - we are using OF_CONTROL
+#   - we have either OF_SEPARATE or OF_HOSTFILE
+build_dtb :=
+ifeq ($(CONFIG_$(SPL_TPL_)OF_PLATDATA),)
+ifneq ($(CONFIG_$(SPL_TPL_)OF_CONTROL),)
+ifeq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_HOSTFILE),y)
+build_dtb := y
+endif
+endif
+endif
 
-ifeq ($(CONFIG_$(SPL_TPL_)OF_CONTROL)$(CONFIG_OF_SEPARATE)$(CONFIG_$(SPL_TPL_)OF_PLATDATA),yy)
+ifneq ($(build_dtb),)
 $(obj)/$(SPL_BIN)-dtb.bin: $(obj)/$(SPL_BIN)-nodtb.bin \
 		$(if $(CONFIG_SPL_SEPARATE_BSS),,$(obj)/$(SPL_BIN)-pad.bin) \
 		$(FINAL_DTB_CONTAINER)  FORCE
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#14 lore raw

PATCH v2 6/50 spl: Allow sandbox to build a device-tree file

Quoted 19 lines · Simon Glass

At present only OF_SEPARATE is considered valid for building a device-tree file in SPL. However sandbox uses OF_HOSTFILE instead. Update the logic to handle this and make it easier to understand.

Note that the new logic is not quite the same as the old logic. It was previously assumed that checking for:

   $(CONFIG_$(SPL_TPL_)OF_CONTROL)
   $(CONFIG_OF_SEPARATE)
   $(CONFIG_$(SPL_TPL_)OF_PLATDATA)

producing 'yy' meant that the first two were 'y' and the last was empty. Strictly speaking it would be possible for any two of the three to be 'y' and still yield the same result. However, that was not the intention of the new logic, since OF_PLATDATA always ensures that no device-tree file is included. So in effect the new logic is the same, with the addition of OF_HOSTFILE as an option for OP_SEPARATE.

Quoted 9 lines

---

Changes in v2:
- Add a better explanation of the logic change, in the commit message

 scripts/Makefile.spl | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
#15 lore raw

PATCH v2 6/50 spl: Allow sandbox to build a device-tree file

Quoted 24 lines · Bin Meng
Quoted 22 lines · Simon Glass

At present only OF_SEPARATE is considered valid for building a device-tree file in SPL. However sandbox uses OF_HOSTFILE instead. Update the logic to handle this and make it easier to understand.

Note that the new logic is not quite the same as the old logic. It was previously assumed that checking for:

   $(CONFIG_$(SPL_TPL_)OF_CONTROL)
   $(CONFIG_OF_SEPARATE)
   $(CONFIG_$(SPL_TPL_)OF_PLATDATA)

producing 'yy' meant that the first two were 'y' and the last was empty. Strictly speaking it would be possible for any two of the three to be 'y' and still yield the same result. However, that was not the intention of the new logic, since OF_PLATDATA always ensures that no device-tree file is included. So in effect the new logic is the same, with the addition of OF_HOSTFILE as an option for OP_SEPARATE.

---

Quoted 8 lines
Quoted 6 lines

Changes in v2:
- Add a better explanation of the logic change, in the commit message

 scripts/Makefile.spl | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

applied to u-boot-x86, thanks!

#16 lore raw

PATCH v2 7/50 binman: Allow sections to have an offset

In reply to Simon Glass

At present sections are always placed automatically. Even if an 'offset' property is provided it is ignored. Update the logic to support an offset for sections.

---

Changes in v2:
- Fix map output when section offset is not set (make it 0)
- Add a test for sections with offsets

 tools/binman/README                       |  7 +++++
 tools/binman/bsection.py                  |  9 +++---
 tools/binman/etype/section.py             |  3 +-
 tools/binman/ftest.py                     | 18 ++++++++++++
 tools/binman/test/101_sections_offset.dts | 35 +++++++++++++++++++++++
 5 files changed, 67 insertions(+), 5 deletions(-)
 create mode 100644 tools/binman/test/101_sections_offset.dts
tools/binman/README +7 -0
index 04ed2b799c8..927fa856acf 100644
@@ -342,6 +342,13 @@ size:
 	Sets the image size in bytes, for example 'size = <0x100000>' for a
 	1MB image.
 
+offset:
+	This is similar to 'offset' in entries, setting the offset of a section
+	within the image or section containing it. The first byte of the section
+	is normally at offset 0. If 'offset' is not provided, binman sets it to
+	the end of the previous region, or the start of the image's entry area
+	(normally 0) if there is no previous region.
+
 align-size:
 	This sets the alignment of the image size. For example, to ensure
 	that the image ends on a 512-byte boundary, use 'align-size = <512>'.
tools/binman/bsection.py +5 -4
index ccf2920c5bd..0ba542ee987 100644
@@ -57,7 +57,7 @@ class Section(object):
         self._name = name
         self._node = node
         self._image = image
-        self._offset = 0
+        self._offset = None
         self._size = None
         self._align_size = None
         self._pad_before = 0
@@ -75,6 +75,7 @@ class Section(object):
 
     def _ReadNode(self):
         """Read properties from the section node"""
+        self._offset = fdt_util.GetInt(self._node, 'offset')
         self._size = fdt_util.GetInt(self._node, 'size')
         self._align_size = fdt_util.GetInt(self._node, 'align-size')
         if tools.NotPowerOfTwo(self._align_size):
@@ -130,7 +131,7 @@ class Section(object):
             entry.AddMissingProperties()
 
     def SetCalculatedProperties(self):
-        state.SetInt(self._node, 'offset', self._offset)
+        state.SetInt(self._node, 'offset', self._offset or 0)
         state.SetInt(self._node, 'size', self._size)
         image_pos = self._image_pos
         if self._parent_section:
@@ -424,8 +425,8 @@ class Section(object):
         Args:
             fd: File to write the map to
         """
-        Entry.WriteMapLine(fd, indent, self._name, self._offset, self._size,
-                           self._image_pos)
+        Entry.WriteMapLine(fd, indent, self._name, self._offset or 0,
+                           self._size, self._image_pos)
         for entry in self._entries.values():
             entry.WriteMap(fd, indent + 1)
 
tools/binman/etype/section.py +2 -1
index 7f1b4136049..3681a484689 100644
@@ -67,7 +67,8 @@ class Entry_section(Entry):
     def Pack(self, offset):
         """Pack all entries into the section"""
         self._section.PackEntries()
-        self._section.SetOffset(offset)
+        if self._section._offset is None:
+            self._section.SetOffset(offset)
         self.size = self._section.GetSize()
         return super(Entry_section, self).Pack(offset)
 
tools/binman/ftest.py +18 -0
index 75658c9a366..daea1ea1382 100644
@@ -1783,6 +1783,24 @@ class TestFunctional(unittest.TestCase):
         data = self._DoReadFile('100_intel_refcode.dts')
         self.assertEqual(REFCODE_DATA, data[:len(REFCODE_DATA)])
 
+    def testSectionOffset(self):
+        """Tests use of a section with an offset"""
+        data, _, map_data, _ = self._DoReadFileDtb('101_sections_offset.dts',
+                                                   map=True)
+        self.assertEqual('''ImagePos    Offset      Size  Name
+00000000  00000000  00000038  main-section
+00000004   00000004  00000010  section at 0
+00000004    00000000  00000004  u-boot
+00000018   00000018  00000010  section at 1
+00000018    00000000  00000004  u-boot
+0000002c   0000002c  00000004  section at 2
+0000002c    00000000  00000004  u-boot
+''', map_data)
+        self.assertEqual(data,
+                         4 * chr(0x26) + U_BOOT_DATA + 12 * chr(0x21) +
+                         4 * chr(0x26) + U_BOOT_DATA + 12 * chr(0x61) +
+                         4 * chr(0x26) + U_BOOT_DATA + 8 * chr(0x26))
+
 
 if __name__ == "__main__":
     unittest.main()
tools/binman/test/101_sections_offset.dts new file +35 -0
new file mode 100644
index 00000000000..46708ff9b6b
@@ -0,0 +1,35 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/dts-v1/;
+
+/ {
+	#address-cells = <1>;
+	#size-cells = <1>;
+
+	binman {
+		pad-byte = <0x26>;
+		size = <0x38>;
+		section at 0 {
+			read-only;
+			offset = <0x4>;
+			size = <0x10>;
+			pad-byte = <0x21>;
+
+			u-boot {
+			};
+		};
+		section at 1 {
+			size = <0x10>;
+			pad-byte = <0x61>;
+			offset = <0x18>;
+
+			u-boot {
+			};
+		};
+		section at 2 {
+			offset = <0x2c>;
+			u-boot {
+			};
+		};
+	};
+};
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#17 lore raw

PATCH v2 7/50 binman: Allow sections to have an offset

Quoted 12 lines · Simon Glass

At present sections are always placed automatically. Even if an 'offset' property is provided it is ignored. Update the logic to support an offset for sections.

---

Changes in v2:
- Fix map output when section offset is not set (make it 0)
- Add a test for sections with offsets

Quoted 8 lines
 tools/binman/README                       |  7 +++++
 tools/binman/bsection.py                  |  9 +++---
 tools/binman/etype/section.py             |  3 +-
 tools/binman/ftest.py                     | 18 ++++++++++++
 tools/binman/test/101_sections_offset.dts | 35 +++++++++++++++++++++++
 5 files changed, 67 insertions(+), 5 deletions(-)
 create mode 100644 tools/binman/test/101_sections_offset.dts
#18 lore raw

PATCH v2 7/50 binman: Allow sections to have an offset

Quoted 23 lines · Bin Meng
Quoted 12 lines · Simon Glass

At present sections are always placed automatically. Even if an 'offset' property is provided it is ignored. Update the logic to support an offset for sections.

---

Changes in v2:
- Fix map output when section offset is not set (make it 0)
- Add a test for sections with offsets

Quoted 8 lines
 tools/binman/README                       |  7 +++++
 tools/binman/bsection.py                  |  9 +++---
 tools/binman/etype/section.py             |  3 +-
 tools/binman/ftest.py                     | 18 ++++++++++++
 tools/binman/test/101_sections_offset.dts | 35 +++++++++++++++++++++++
 5 files changed, 67 insertions(+), 5 deletions(-)
 create mode 100644 tools/binman/test/101_sections_offset.dts
Quoted 1 line

applied to u-boot-x86, thanks!

#19 lore raw

PATCH v2 8/50 x86: start64: Fix copyright message

In reply to Simon Glass

There is a typo in this header. Fix it.

---

Changes in v2: None

 arch/x86/cpu/start64.S | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/cpu/start64.S +1 -1
index a78a3316b6d..7be834788b9 100644
@@ -2,7 +2,7 @@
 /*
  * 64-bit x86 Startup Code
  *
- * (C) Copyright 216 Google, Inc
+ * Copyright 2019 Google, Inc
  * Written by Simon Glass <[email protected]>
  */
 
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#21 lore raw

PATCH v2 9/50 x86: mp_init: Use proper error numbers

In reply to Simon Glass

At present many of the functions in this file return -1 as an error number. which is -EPERM. Update the code to use real error numbers.

---

Changes in v2: None

 arch/x86/cpu/mp_init.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
arch/x86/cpu/mp_init.c +5 -5
index ea64c2ee57e..fefbf8f7282 100644
@@ -322,7 +322,7 @@ static int start_aps(int ap_count, atomic_t *num_aps)
 	if (sipi_vector > max_vector_loc) {
 		printf("SIPI vector too large! 0x%08x\n",
 		       sipi_vector);
-		return -1;
+		return -ENOSPC;
 	}
 
 	debug("Attempting to start %d APs\n", ap_count);
@@ -364,7 +364,7 @@ static int start_aps(int ap_count, atomic_t *num_aps)
 	if (wait_for_aps(num_aps, ap_count, 10000, 50)) {
 		debug("Not all APs checked in: %d/%d\n",
 		      atomic_read(num_aps), ap_count);
-		return -1;
+		return -EIO;
 	}
 
 	return 0;
@@ -387,7 +387,7 @@ static int bsp_do_flight_plan(struct udevice *cpu, struct mp_params *mp_params)
 			if (wait_for_aps(&rec->cpus_entered, num_aps,
 					 timeout_us, step_us)) {
 				debug("MP record %d timeout\n", i);
-				ret = -1;
+				ret = -ETIMEDOUT;
 			}
 		}
 
@@ -508,7 +508,7 @@ int mp_init(struct mp_params *p)
 
 	if (p == NULL || p->flight_plan == NULL || p->num_records < 1) {
 		printf("Invalid MP parameters\n");
-		return -1;
+		return -EINVAL;
 	}
 
 	num_cpus = cpu_get_count(cpu);
@@ -531,7 +531,7 @@ int mp_init(struct mp_params *p)
 	/* Load the SIPI vector */
 	ret = load_sipi_vector(&ap_count, num_cpus);
 	if (ap_count == NULL)
-		return -1;
+		return -ENOENT;
 
 	/*
 	 * Make sure SIPI data hits RAM so the APs that come up will see
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#22 lore raw

PATCH v2 9/50 x86: mp_init: Use proper error numbers

Quoted 4 lines · Simon Glass

At present many of the functions in this file return -1 as an error number. which is -EPERM. Update the code to use real error numbers.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/cpu/mp_init.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

applied to u-boot-x86, thanks!

#23 lore raw

PATCH v2 10/50 x86: Add a way to reinit the cpu

In reply to Simon Glass

We cannot init the CPU fully both than once during a boot. Add a new function which can be called to figure out the CPU identity, but which does not change anything. For x86_64, this is empty for now.

---

Changes in v2: None

 arch/x86/cpu/i386/cpu.c           | 113 ++++++++++++++++++------------
 arch/x86/cpu/x86_64/cpu.c         |   5 ++
 arch/x86/include/asm/u-boot-x86.h |  20 ++++++
 3 files changed, 94 insertions(+), 44 deletions(-)
arch/x86/cpu/i386/cpu.c +69 -44
index 3bde44ebf53..90b546e7410 100644
@@ -309,21 +309,22 @@ u32 cpu_get_stepping(void)
 	return gd->arch.x86_mask;
 }
 
-int x86_cpu_init_f(void)
+/* initialise FPU, reset EM, set MP and NE */
+static void setup_cpu_features(void)
 {
 	const u32 em_rst = ~X86_CR0_EM;
 	const u32 mp_ne_set = X86_CR0_MP | X86_CR0_NE;
 
-	if (ll_boot_init()) {
-		/* initialize FPU, reset EM, set MP and NE */
-		asm ("fninit\n" \
-		"movl %%cr0, %%eax\n" \
-		"andl %0, %%eax\n" \
-		"orl  %1, %%eax\n" \
-		"movl %%eax, %%cr0\n" \
-		: : "i" (em_rst), "i" (mp_ne_set) : "eax");
-	}
+	asm ("fninit\n" \
+	"movl %%cr0, %%eax\n" \
+	"andl %0, %%eax\n" \
+	"orl  %1, %%eax\n" \
+	"movl %%eax, %%cr0\n" \
+	: : "i" (em_rst), "i" (mp_ne_set) : "eax");
+}
 
+static void setup_identity(void)
+{
 	/* identify CPU via cpuid and store the decoded info into gd->arch */
 	if (has_cpuid()) {
 		struct cpu_device_id cpu;
@@ -339,46 +340,70 @@ int x86_cpu_init_f(void)
 
 		gd->arch.has_mtrr = has_mtrr();
 	}
-	/* Don't allow PCI region 3 to use memory in the 2-4GB memory hole */
+}
+
+/* Don't allow PCI region 3 to use memory in the 2-4GB memory hole */
+static void setup_pci_ram_top(void)
+{
 	gd->pci_ram_top = 0x80000000U;
+}
+
+static void setup_mtrr(void)
+{
+	u64 mtrr_cap;
 
 	/* Configure fixed range MTRRs for some legacy regions */
-	if (gd->arch.has_mtrr) {
-		u64 mtrr_cap;
-
-		mtrr_cap = native_read_msr(MTRR_CAP_MSR);
-		if (mtrr_cap & MTRR_CAP_FIX) {
-			/* Mark the VGA RAM area as uncacheable */
-			native_write_msr(MTRR_FIX_16K_A0000_MSR,
-					 MTRR_FIX_TYPE(MTRR_TYPE_UNCACHEABLE),
-					 MTRR_FIX_TYPE(MTRR_TYPE_UNCACHEABLE));
-
-			/*
-			 * Mark the PCI ROM area as cacheable to improve ROM
-			 * execution performance.
-			 */
-			native_write_msr(MTRR_FIX_4K_C0000_MSR,
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
-			native_write_msr(MTRR_FIX_4K_C8000_MSR,
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
-			native_write_msr(MTRR_FIX_4K_D0000_MSR,
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
-			native_write_msr(MTRR_FIX_4K_D8000_MSR,
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
-					 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
-
-			/* Enable the fixed range MTRRs */
-			msr_setbits_64(MTRR_DEF_TYPE_MSR, MTRR_DEF_TYPE_FIX_EN);
-		}
+	if (!gd->arch.has_mtrr)
+		return;
+
+	mtrr_cap = native_read_msr(MTRR_CAP_MSR);
+	if (mtrr_cap & MTRR_CAP_FIX) {
+		/* Mark the VGA RAM area as uncacheable */
+		native_write_msr(MTRR_FIX_16K_A0000_MSR,
+				 MTRR_FIX_TYPE(MTRR_TYPE_UNCACHEABLE),
+				 MTRR_FIX_TYPE(MTRR_TYPE_UNCACHEABLE));
+
+		/*
+		 * Mark the PCI ROM area as cacheable to improve ROM
+		 * execution performance.
+		 */
+		native_write_msr(MTRR_FIX_4K_C0000_MSR,
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
+		native_write_msr(MTRR_FIX_4K_C8000_MSR,
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
+		native_write_msr(MTRR_FIX_4K_D0000_MSR,
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
+		native_write_msr(MTRR_FIX_4K_D8000_MSR,
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK),
+				 MTRR_FIX_TYPE(MTRR_TYPE_WRBACK));
+
+		/* Enable the fixed range MTRRs */
+		msr_setbits_64(MTRR_DEF_TYPE_MSR, MTRR_DEF_TYPE_FIX_EN);
 	}
+}
+
+int x86_cpu_init_f(void)
+{
+	if (ll_boot_init())
+		setup_cpu_features();
+	setup_identity();
+	setup_mtrr();
+	setup_pci_ram_top();
 
-#ifdef CONFIG_I8254_TIMER
 	/* Set up the i8254 timer if required */
-	i8254_init();
-#endif
+	if (IS_ENABLED(CONFIG_I8254_TIMER))
+		i8254_init();
+
+	return 0;
+}
+
+int x86_cpu_reinit_f(void)
+{
+	setup_identity();
+	setup_pci_ram_top();
 
 	return 0;
 }
arch/x86/cpu/x86_64/cpu.c +5 -0
index 6c063e82009..42abb23a9ed 100644
@@ -61,3 +61,8 @@ int print_cpuinfo(void)
 {
 	return 0;
 }
+
+int x86_cpu_reinit_f(void)
+{
+	return 0;
+}
arch/x86/include/asm/u-boot-x86.h +20 -0
index 670fcdc0093..c252192bf41 100644
@@ -13,7 +13,27 @@ extern char gdt_rom[];
 
 /* cpu/.../cpu.c */
 int arch_cpu_init(void);
+
+/**
+ * x86_cpu_init_f() - Set up basic features of the x86 CPU
+ *
+ * 0 on success, -ve on error
+ */
 int x86_cpu_init_f(void);
+
+/**
+ * x86_cpu_reinit_f() - Set up the CPU a second time
+ *
+ * Once cpu_init_f() has been called (e.g. in SPL) we should not call it
+ * again (e.g. in U-Boot proper) since it sets up the state from scratch.
+ * Call this function in later phases of U-Boot instead. It reads the CPU
+ * identify so that CPU functions can be used correctly, but does not change
+ * anything.
+ *
+ * @return 0 (indicating success, to mimic cpu_init_f())
+ */
+int x86_cpu_reinit_f(void);
+
 int cpu_init_f(void);
 void setup_gdt(struct global_data *id, u64 *gdt_addr);
 /*
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#24 lore raw

PATCH v2 10/50 x86: Add a way to reinit the cpu

Quoted 7 lines · Simon Glass

We cannot init the CPU fully both than once during a boot. Add a new function which can be called to figure out the CPU identity, but which does not change anything. For x86_64, this is empty for now.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/cpu/i386/cpu.c           | 113 ++++++++++++++++++------------
 arch/x86/cpu/x86_64/cpu.c         |   5 ++
 arch/x86/include/asm/u-boot-x86.h |  20 ++++++
 3 files changed, 94 insertions(+), 44 deletions(-)

applied to u-boot-x86, thanks!

#25 lore raw

PATCH v2 11/50 x86: dts: Add device-tree labels for rtc and reset

In reply to Simon Glass

Add labels for these nodes so that board DT files can reference them.

---

Changes in v2: None

 arch/x86/dts/reset.dtsi | 2 +-
 arch/x86/dts/rtc.dtsi   | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
arch/x86/dts/reset.dtsi +1 -1
index f979d83757c..555d0dd9608 100644
@@ -1,5 +1,5 @@
 / {
-	reset {
+	reset: reset {
 		compatible = "x86,reset";
 		u-boot,dm-pre-reloc;
 	};
arch/x86/dts/rtc.dtsi +1 -1
index 1797e042daf..d0bbd84e509 100644
@@ -1,5 +1,5 @@
 / {
-	rtc {
+	rtc: rtc {
 		compatible = "motorola,mc146818";
 		u-boot,dm-pre-reloc;
 		reg = <0x70 2>;
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#26 lore raw

PATCH v2 11/50 x86: dts: Add device-tree labels for rtc and reset

Quoted 5 lines · Simon Glass

Add labels for these nodes so that board DT files can reference them.

Quoted 8 lines

---

Changes in v2: None

 arch/x86/dts/reset.dtsi | 2 +-
 arch/x86/dts/rtc.dtsi   | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

applied to u-boot-x86, thanks!

#27 lore raw

PATCH v2 12/50 x86: Update a stale comment about ifdtool

In reply to Simon Glass

We use binman to build the x86 image now. Update a comment which still refers to ifdtool.

---

Changes in v2:
- Update the comment in fsp_cap.S too

 arch/x86/cpu/intel_common/car.S | 2 +-
 arch/x86/lib/fsp/fsp_car.S      | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
arch/x86/cpu/intel_common/car.S +1 -1
index 52a77bb2d18..00308dbdef9 100644
@@ -235,7 +235,7 @@ mtrr_table_end:
 
 	.align 4
 _dt_ucode_base_size:
-	/* These next two fields are filled in by ifdtool */
+	/* These next two fields are filled in by binman */
 .globl ucode_base
 ucode_base:	/* Declared in microcode.h */
 	.long	0			/* microcode base */
arch/x86/lib/fsp/fsp_car.S +1 -1
index 48edc8362a7..8c54cea3db4 100644
@@ -100,7 +100,7 @@ temp_ram_init_romstack:
 	.long	temp_ram_init_params
 temp_ram_init_params:
 _dt_ucode_base_size:
-	/* These next two fields are filled in by ifdtool */
+	/* These next two fields are filled in by binman */
 .globl ucode_base
 ucode_base:	/* Declared in microcode.h */
 	.long	0			/* microcode base */
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#28 lore raw

PATCH v2 12/50 x86: Update a stale comment about ifdtool

Quoted 6 lines · Simon Glass

We use binman to build the x86 image now. Update a comment which still refers to ifdtool.

Quoted 9 lines

---

Changes in v2:
- Update the comment in fsp_cap.S too

 arch/x86/cpu/intel_common/car.S | 2 +-
 arch/x86/lib/fsp/fsp_car.S      | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

applied to u-boot-x86, thanks!

#29 lore raw

PATCH v2 13/50 x86: Support SPL and TPL

In reply to Simon Glass

At present only chromebook_link64 supports SPL. It is useful to eb able to support both TPL and SPL to implement verified boot on x86.

Enable the options for both along with some suitable default options needed to boot through these phases.

---

Changes in v2: None

 arch/Kconfig     | 31 +++++++++++++++++++++++++++++++
 arch/x86/Kconfig |  1 -
 2 files changed, 31 insertions(+), 1 deletion(-)
arch/Kconfig +31 -0
index 2f3d07c13a1..a1d1ac301d6 100644
@@ -126,6 +126,8 @@ config SH
 
 config X86
 	bool "x86 architecture"
+	select SUPPORT_SPL
+	select SUPPORT_TPL
 	select CREATE_ARCH_SYMLINK
 	select DM
 	select DM_PCI
@@ -164,6 +166,35 @@ config X86
 	imply USB_HOST_ETHER
 	imply PCH
 
+	# Thing to enable for when SPL/TPL are enabled: SPL
+	imply SPL_DM
+	imply SPL_OF_LIBFDT
+	imply SPL_DRIVERS_MISC_SUPPORT
+	imply SPL_GPIO_SUPPORT
+	imply SPL_LIBCOMMON_SUPPORT
+	imply SPL_LIBGENERIC_SUPPORT
+	imply SPL_SERIAL_SUPPORT
+	imply SPL_SPI_FLASH_SUPPORT
+	imply SPL_SPI_SUPPORT
+	imply SPL_OF_CONTROL
+	imply SPL_TIMER
+	imply SPL_REGMAP
+	imply SPL_SYSCON
+	# TPL
+	imply TPL_DM
+	imply TPL_OF_LIBFDT
+	imply TPL_DRIVERS_MISC_SUPPORT
+	imply TPL_GPIO_SUPPORT
+	imply TPL_LIBCOMMON_SUPPORT
+	imply TPL_LIBGENERIC_SUPPORT
+	imply TPL_SERIAL_SUPPORT
+	imply TPL_SPI_FLASH_SUPPORT
+	imply TPL_SPI_SUPPORT
+	imply TPL_OF_CONTROL
+	imply TPL_TIMER
+	imply TPL_REGMAP
+	imply TPL_SYSCON
+
 config XTENSA
 	bool "Xtensa architecture"
 	select CREATE_ARCH_SYMLINK
arch/x86/Kconfig +0 -1
index e0520937751..185f0ef8c40 100644
@@ -32,7 +32,6 @@ config X86_RUN_32BIT
 config X86_RUN_64BIT
 	bool "64-bit"
 	select X86_64
-	select SUPPORT_SPL
 	select SPL
 	select SPL_SEPARATE_BSS
 	help
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#30 lore raw

PATCH v2 13/50 x86: Support SPL and TPL

Quoted 9 lines · Simon Glass

At present only chromebook_link64 supports SPL. It is useful to eb able to support both TPL and SPL to implement verified boot on x86.

Enable the options for both along with some suitable default options needed to boot through these phases.

Quoted 8 lines

---

Changes in v2: None

 arch/Kconfig     | 31 +++++++++++++++++++++++++++++++
 arch/x86/Kconfig |  1 -
 2 files changed, 31 insertions(+), 1 deletion(-)

applied to u-boot-x86, thanks!

#31 lore raw

PATCH v2 14/50 x86: Support booting with TPL

In reply to Simon Glass

Some boards want to use TPL as the first phase of U-Boot. This allows selection of A or B SPL phases, thus allowing the memory init to be upgraded in the field.

Add a new Kconfig option for this.

---

Changes in v2: None

 arch/x86/Kconfig | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
arch/x86/Kconfig +8 -1
index 185f0ef8c40..45a533625a6 100644
@@ -176,10 +176,17 @@ config X86_16BIT_INIT
 config SPL_X86_16BIT_INIT
 	bool
 	depends on X86_RESET_VECTOR
-	default y if X86_RESET_VECTOR && SPL
+	default y if X86_RESET_VECTOR && SPL && !TPL
 	help
 	  This is enabled when 16-bit init is in SPL
 
+config TPL_X86_16BIT_INIT
+	bool
+	depends on X86_RESET_VECTOR
+	default y if X86_RESET_VECTOR && TPL
+	help
+	  This is enabled when 16-bit init is in TPL
+
 config X86_32BIT_INIT
 	bool
 	depends on X86_RESET_VECTOR
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#32 lore raw

PATCH v2 14/50 x86: Support booting with TPL

Quoted 7 lines · Simon Glass

Some boards want to use TPL as the first phase of U-Boot. This allows selection of A or B SPL phases, thus allowing the memory init to be upgraded in the field.

Add a new Kconfig option for this.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/Kconfig | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

applied to u-boot-x86, thanks!

#33 lore raw

PATCH v2 15/50 x86: Add a handoff header file

In reply to Simon Glass

Add an arch-specific handoff header so that we can use the HANDOFF feature on x86 devices.

---

Changes in v2: None

 arch/x86/include/asm/handoff.h | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
 create mode 100644 arch/x86/include/asm/handoff.h
arch/x86/include/asm/handoff.h new file +15 -0
new file mode 100644
index 00000000000..4d18d59efed
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Architecture-specific SPL handoff information for x86
+ *
+ * Copyright 2018 Google, Inc
+ * Written by Simon Glass <[email protected]>
+ */
+
+#ifndef __x86_asm_handoff_h
+#define __x86_asm_handoff_h
+
+struct arch_spl_handoff {
+};
+
+#endif
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#34 lore raw

PATCH v2 15/50 x86: Add a handoff header file

Quoted 6 lines · Simon Glass

Add an arch-specific handoff header so that we can use the HANDOFF feature on x86 devices.

Quoted 8 lines

---

Changes in v2: None

 arch/x86/include/asm/handoff.h | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
 create mode 100644 arch/x86/include/asm/handoff.h

applied to u-boot-x86, thanks!

#35 lore raw

PATCH v2 16/50 x86: broadwell: Improve SDRAM debugging output

In reply to Simon Glass

Add debugging during SDRAM init so that problems are easier to diagnose.

---

Changes in v2: None

 arch/x86/cpu/broadwell/sdram.c | 32 ++++++++++++++++++++------------
 1 file changed, 20 insertions(+), 12 deletions(-)
arch/x86/cpu/broadwell/sdram.c +20 -12
index 03a35bcf73f..1b9f9840c6f 100644
@@ -204,16 +204,18 @@ int dram_init(void)
 
 	/* Print ME state before MRC */
 	ret = syscon_get_by_driver_data(X86_SYSCON_ME, &me_dev);
-	if (ret)
+	if (ret) {
+		debug("Cannot get ME (err=%d)\n", ret);
 		return ret;
+	}
 	intel_me_status(me_dev);
 
 	/* Save ME HSIO version */
-	ret = uclass_first_device(UCLASS_PCH, &pch_dev);
-	if (ret)
+	ret = uclass_first_device_err(UCLASS_PCH, &pch_dev);
+	if (ret) {
+		debug("Cannot get PCH (err=%d)\n", ret);
 		return ret;
-	if (!pch_dev)
-		return -ENODEV;
+	}
 	power_state_get(pch_dev, &ps);
 
 	intel_me_hsio_version(me_dev, &ps.hsio_version, &ps.hsio_checksum);
@@ -221,15 +223,17 @@ int dram_init(void)
 	broadwell_fill_pei_data(pei_data);
 	mainboard_fill_pei_data(pei_data);
 
-	ret = uclass_first_device(UCLASS_NORTHBRIDGE, &dev);
-	if (ret)
+	ret = uclass_first_device_err(UCLASS_NORTHBRIDGE, &dev);
+	if (ret) {
+		debug("Cannot get Northbridge (err=%d)\n", ret);
 		return ret;
-	if (!dev)
-		return -ENODEV;
+	}
 	size = 256;
 	ret = mrc_locate_spd(dev, size, &spd_data);
-	if (ret)
+	if (ret) {
+		debug("Cannot locate SPD (err=%d)\n", ret);
 		return ret;
+	}
 	memcpy(pei_data->spd_data[0][0], spd_data, size);
 	memcpy(pei_data->spd_data[1][0], spd_data, size);
 
@@ -239,13 +243,17 @@ int dram_init(void)
 
 	debug("PEI version %#x\n", pei_data->pei_version);
 	ret = mrc_common_init(dev, pei_data, true);
-	if (ret)
+	if (ret) {
+		debug("mrc_common_init() failed(err=%d)\n", ret);
 		return ret;
+	}
 	debug("Memory init done\n");
 
 	ret = sdram_find(dev);
-	if (ret)
+	if (ret) {
+		debug("sdram_find() failed (err=%d)\n", ret);
 		return ret;
+	}
 	gd->ram_size = gd->arch.meminfo.total_32bit_memory;
 	debug("RAM size %llx\n", (unsigned long long)gd->ram_size);
 
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#36 lore raw

PATCH v2 16/50 x86: broadwell: Improve SDRAM debugging output

Quoted 4 lines · Simon Glass

Add debugging during SDRAM init so that problems are easier to diagnose.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/cpu/broadwell/sdram.c | 32 ++++++++++++++++++++------------
 1 file changed, 20 insertions(+), 12 deletions(-)

applied to u-boot-x86, thanks!

#37 lore raw

PATCH v2 17/50 x86: broadwell: Allow SDRAM init from SPL

In reply to Simon Glass

At present, for broadwell, SDRAM is always set up in U-Boot proper since the 64-bit mode (which uses SDRAM init in SPL) is not supported.

Update the code to allow SDRAM init in SPL instead so that U-Boot proper can be loaded into SDRAM and run from there. This allows U-Boot to be compressed to reduce space, since it is not necessary to run it directly from flash. It could later allow us to support 64-bit U-Boot on broadwell.

---

Changes in v2:
- Update commit message to make it clear this patch is just for broadwell
- Bring in sdram_console_tx_byte() to allow debugging

 arch/x86/cpu/broadwell/Makefile      |   2 +-
 arch/x86/cpu/broadwell/northbridge.c | 100 +++++++++++++++++++++++++++
 arch/x86/cpu/broadwell/sdram.c       |  93 -------------------------
 3 files changed, 101 insertions(+), 94 deletions(-)
arch/x86/cpu/broadwell/Makefile +1 -1
index d3785aabdf2..d464507e6c0 100644
@@ -13,4 +13,4 @@ obj-y += pinctrl_broadwell.o
 obj-y += power_state.o
 obj-y += refcode.o
 obj-y += sata.o
-obj-y += sdram.o
+obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += sdram.o
arch/x86/cpu/broadwell/northbridge.c +100 -0

108 lines not on this page · raw message

arch/x86/cpu/broadwell/sdram.c +0 -93

99 lines not on this page · raw message

Signature · 1 line
2.21.0.593.g511ec345e18-goog
#38 lore raw

PATCH v2 17/50 x86: broadwell: Allow SDRAM init from SPL

Quoted 12 lines · Simon Glass

At present, for broadwell, SDRAM is always set up in U-Boot proper since the 64-bit mode (which uses SDRAM init in SPL) is not supported.

Update the code to allow SDRAM init in SPL instead so that U-Boot proper can be loaded into SDRAM and run from there. This allows U-Boot to be compressed to reduce space, since it is not necessary to run it directly from flash. It could later allow us to support 64-bit U-Boot on broadwell.

---

Quoted 9 lines

Changes in v2:
- Update commit message to make it clear this patch is just for broadwell
- Bring in sdram_console_tx_byte() to allow debugging

 arch/x86/cpu/broadwell/Makefile      |   2 +-
 arch/x86/cpu/broadwell/northbridge.c | 100 +++++++++++++++++++++++++++
 arch/x86/cpu/broadwell/sdram.c       |  93 -------------------------
 3 files changed, 101 insertions(+), 94 deletions(-)
#39 lore raw

PATCH v2 17/50 x86: broadwell: Allow SDRAM init from SPL

Quoted 18 lines · Bin Meng
Quoted 16 lines · Simon Glass

At present, for broadwell, SDRAM is always set up in U-Boot proper since the 64-bit mode (which uses SDRAM init in SPL) is not supported.

Update the code to allow SDRAM init in SPL instead so that U-Boot proper can be loaded into SDRAM and run from there. This allows U-Boot to be compressed to reduce space, since it is not necessary to run it directly from flash. It could later allow us to support 64-bit U-Boot on broadwell.

---

Changes in v2:
- Update commit message to make it clear this patch is just for broadwell
- Bring in sdram_console_tx_byte() to allow debugging

Quoted 7 lines
Quoted 5 lines
 arch/x86/cpu/broadwell/Makefile      |   2 +-
 arch/x86/cpu/broadwell/northbridge.c | 100 +++++++++++++++++++++++++++
 arch/x86/cpu/broadwell/sdram.c       |  93 -------------------------
 3 files changed, 101 insertions(+), 94 deletions(-)

applied to u-boot-x86, thanks!

#40 lore raw

PATCH v2 18/50 x86: Move init of debug UART to cpu.c

In reply to Simon Glass

At present the debug UART is set up in sdram.c which is not the best place since it has nothing in particular to do with SDRAM. Since we want to support initing this in SPL too, move it to a common file.

---

Changes in v2: None

 arch/x86/cpu/broadwell/cpu.c   | 13 +++++++++++++
 arch/x86/cpu/broadwell/sdram.c | 11 -----------
 2 files changed, 13 insertions(+), 11 deletions(-)
arch/x86/cpu/broadwell/cpu.c +13 -0
index 232fa40eb53..d53c7b863fb 100644
@@ -12,7 +12,9 @@
 #include <asm/cpu_x86.h>
 #include <asm/cpu_common.h>
 #include <asm/intel_regs.h>
+#include <asm/lpc_common.h>
 #include <asm/msr.h>
+#include <asm/pci.h>
 #include <asm/post.h>
 #include <asm/turbo.h>
 #include <asm/arch/cpu.h>
@@ -156,6 +158,17 @@ int print_cpuinfo(void)
 	return 0;
 }
 
+void board_debug_uart_init(void)
+{
+	struct udevice *bus = NULL;
+
+	/* com1 / com2 decode range */
+	pci_x86_write_config(bus, PCH_DEV_LPC, LPC_IO_DEC, 1 << 4, PCI_SIZE_16);
+
+	pci_x86_write_config(bus, PCH_DEV_LPC, LPC_EN, COMA_LPC_EN,
+			     PCI_SIZE_16);
+}
+
 /*
  * The core 100MHz BLCK is disabled in deeper c-states. One needs to calibrate
  * the 100MHz BCLCK against the 24MHz BLCK to restore the clocks properly
arch/x86/cpu/broadwell/sdram.c +0 -11
index b8450cc9d29..b31d78c092a 100644
@@ -194,17 +194,6 @@ int misc_init_r(void)
 	return 0;
 }
 
-void board_debug_uart_init(void)
-{
-	struct udevice *bus = NULL;
-
-	/* com1 / com2 decode range */
-	pci_x86_write_config(bus, PCH_DEV_LPC, LPC_IO_DEC, 1 << 4, PCI_SIZE_16);
-
-	pci_x86_write_config(bus, PCH_DEV_LPC, LPC_EN, COMA_LPC_EN,
-			     PCI_SIZE_16);
-}
-
 static const struct udevice_id broadwell_syscon_ids[] = {
 	{ .compatible = "intel,me", .data = X86_SYSCON_ME },
 	{ }
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#41 lore raw

PATCH v2 18/50 x86: Move init of debug UART to cpu.c

Quoted 7 lines · Simon Glass

At present the debug UART is set up in sdram.c which is not the best place since it has nothing in particular to do with SDRAM. Since we want to support initing this in SPL too, move it to a common file.

Quoted 8 lines

---

Changes in v2: None

 arch/x86/cpu/broadwell/cpu.c   | 13 +++++++++++++
 arch/x86/cpu/broadwell/sdram.c | 11 -----------
 2 files changed, 13 insertions(+), 11 deletions(-)

applied to u-boot-x86, thanks!

#42 lore raw

PATCH v2 19/50 x86: broadwell: Split CPU init

In reply to Simon Glass

Split the CPU init into two parts - the 'full' init which happens in the first U-Boot phase, and the rest of the init that happens on subsequent stages.

---

Changes in v2: None

 arch/x86/cpu/broadwell/Makefile   |   1 +
 arch/x86/cpu/broadwell/cpu.c      | 673 -----------------------------
 arch/x86/cpu/broadwell/cpu_full.c | 694 ++++++++++++++++++++++++++++++
 3 files changed, 695 insertions(+), 673 deletions(-)
 create mode 100644 arch/x86/cpu/broadwell/cpu_full.c
arch/x86/cpu/broadwell/Makefile +1 -0
index d464507e6c0..8553615f67b 100644
@@ -4,6 +4,7 @@
 
 obj-y += adsp.o
 obj-y += cpu.o
+obj-y += cpu_full.o
 obj-y += iobp.o
 obj-y += lpc.o
 obj-y += me.o
arch/x86/cpu/broadwell/cpu.c +0 -673

682 lines not on this page · raw message

arch/x86/cpu/broadwell/cpu_full.c new file +694 -0

694 lines not on this page · raw message

Signature · 1 line
2.21.0.593.g511ec345e18-goog
#43 lore raw

PATCH v2 19/50 x86: broadwell: Split CPU init

Quoted 9 lines · Simon Glass

Split the CPU init into two parts - the 'full' init which happens in the first U-Boot phase, and the rest of the init that happens on subsequent stages.

---

Quoted 8 lines

Changes in v2: None

 arch/x86/cpu/broadwell/Makefile   |   1 +
 arch/x86/cpu/broadwell/cpu.c      | 673 -----------------------------
 arch/x86/cpu/broadwell/cpu_full.c | 694 ++++++++++++++++++++++++++++++
 3 files changed, 695 insertions(+), 673 deletions(-)
 create mode 100644 arch/x86/cpu/broadwell/cpu_full.c

applied to u-boot-x86, thanks!

#44 lore raw

PATCH v2 20/50 x86: Add support for starting from SPL/TPL

In reply to Simon Glass

When a previous phase of U-Boot has run we need to adjust the init of subsequent states to avoid messing up the CPU state.

Add a new version of the start logic for SPL, when it boots from TPL (start_from tpl.c) and a new version for U-Boot when it boots from SPL.

---

Changes in v2:
- Add xorl to TPL code also
- Update comments in start_from_tpl to correctly explain SPL state

 arch/x86/Makefile             | 12 ++++++
 arch/x86/cpu/Makefile         | 15 +++++++-
 arch/x86/cpu/start_from_spl.S | 71 +++++++++++++++++++++++++++++++++++
 arch/x86/cpu/start_from_tpl.S | 49 ++++++++++++++++++++++++
 4 files changed, 146 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/cpu/start_from_spl.S
 create mode 100644 arch/x86/cpu/start_from_tpl.S
arch/x86/Makefile +12 -0
index fec14847cc5..9fd6cf2d3bb 100644
@@ -4,9 +4,21 @@ ifeq ($(CONFIG_EFI_APP),)
 ifdef CONFIG_$(SPL_)X86_64
 head-y := arch/x86/cpu/start64.o
 else
+ifeq ($(CONFIG_$(SPL_TPL_)X86_16BIT_INIT),y)
 head-y := arch/x86/cpu/start.o
+else
+ifndef CONFIG_SPL
+head-y := arch/x86/cpu/start.o
+else
+ifdef CONFIG_SPL_BUILD
+head-y	= arch/x86/cpu/start_from_tpl.o
+else
+head-y	= arch/x86/cpu/start_from_spl.o
+endif
+endif
 endif
 endif
+endif # EFI
 
 head-$(CONFIG_$(SPL_)X86_16BIT_INIT) += arch/x86/cpu/start16.o
 head-$(CONFIG_$(SPL_)X86_16BIT_INIT) += arch/x86/cpu/resetvec.o
arch/x86/cpu/Makefile +14 -1
index 54668aab240..85fd5e616ea 100644
@@ -9,9 +9,22 @@
 ifeq ($(CONFIG_$(SPL_)X86_64),y)
 extra-y	= start64.o
 else
+ifeq ($(CONFIG_$(SPL_TPL_)X86_16BIT_INIT),y)
 extra-y	= start.o
+else
+ifndef CONFIG_SPL
+extra-y	= start.o
+else
+ifdef CONFIG_SPL_BUILD
+extra-y	= start_from_tpl.o
+else
+extra-y	= start_from_spl.o
 endif
-extra-$(CONFIG_$(SPL_)X86_16BIT_INIT) += resetvec.o start16.o
+endif
+endif
+endif
+
+extra-$(CONFIG_$(SPL_TPL_)X86_16BIT_INIT) += resetvec.o start16.o
 
 obj-y	+= cpu.o cpu_x86.o
 
arch/x86/cpu/start_from_spl.S new file +71 -0

71 lines not on this page · raw message

arch/x86/cpu/start_from_tpl.S new file +49 -0

49 lines not on this page · raw message

Signature · 1 line
2.21.0.593.g511ec345e18-goog
#45 lore raw

PATCH v2 20/50 x86: Add support for starting from SPL/TPL

Quoted 14 lines · Simon Glass

When a previous phase of U-Boot has run we need to adjust the init of subsequent states to avoid messing up the CPU state.

Add a new version of the start logic for SPL, when it boots from TPL (start_from tpl.c) and a new version for U-Boot when it boots from SPL.

---

Changes in v2:
- Add xorl to TPL code also
- Update comments in start_from_tpl to correctly explain SPL state

Quoted 8 lines
 arch/x86/Makefile             | 12 ++++++
 arch/x86/cpu/Makefile         | 15 +++++++-
 arch/x86/cpu/start_from_spl.S | 71 +++++++++++++++++++++++++++++++++++
 arch/x86/cpu/start_from_tpl.S | 49 ++++++++++++++++++++++++
 4 files changed, 146 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/cpu/start_from_spl.S
 create mode 100644 arch/x86/cpu/start_from_tpl.S
#46 lore raw

PATCH v2 20/50 x86: Add support for starting from SPL/TPL

Quoted 25 lines · Bin Meng
Quoted 14 lines · Simon Glass

When a previous phase of U-Boot has run we need to adjust the init of subsequent states to avoid messing up the CPU state.

Add a new version of the start logic for SPL, when it boots from TPL (start_from tpl.c) and a new version for U-Boot when it boots from SPL.

---

Changes in v2:
- Add xorl to TPL code also
- Update comments in start_from_tpl to correctly explain SPL state

Quoted 8 lines
 arch/x86/Makefile             | 12 ++++++
 arch/x86/cpu/Makefile         | 15 +++++++-
 arch/x86/cpu/start_from_spl.S | 71 +++++++++++++++++++++++++++++++++++
 arch/x86/cpu/start_from_tpl.S | 49 ++++++++++++++++++++++++
 4 files changed, 146 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/cpu/start_from_spl.S
 create mode 100644 arch/x86/cpu/start_from_tpl.S
Quoted 1 line

applied to u-boot-x86, thanks!

#47 lore raw

PATCH v2 21/50 x86: Allow 16-bit init to be in TPL

In reply to Simon Glass

At present we support having 16-bit init be in SPL or U-Boot proper, but not TPL. Add support for this so that TPL can boot.

---

Changes in v2: None

 Makefile                           |  1 +
 arch/x86/Makefile                  |  4 ++--
 arch/x86/cpu/intel_common/Makefile |  2 +-
 arch/x86/cpu/u-boot-spl.lds        |  2 +-
 scripts/Makefile.spl               | 10 +++++++++-
 5 files changed, 14 insertions(+), 5 deletions(-)
Makefile +1 -0
index f2c7bb6041a..56497e4afe5 100644
@@ -1380,6 +1380,7 @@ cmd_ldr = $(LD) $(LDFLAGS_$(@F)) \
 
 u-boot.rom: u-boot-x86-16bit.bin u-boot.bin \
 		$(if $(CONFIG_SPL_X86_16BIT_INIT),spl/u-boot-spl.bin) \
+		$(if $(CONFIG_TPL_X86_16BIT_INIT),tpl/u-boot-tpl.bin) \
 		$(if $(CONFIG_HAVE_REFCODE),refcode.bin) FORCE
 	$(call if_changed,binman)
 
arch/x86/Makefile +2 -2
index 9fd6cf2d3bb..f1afc74fff8 100644
@@ -20,8 +20,8 @@ endif
 endif
 endif # EFI
 
-head-$(CONFIG_$(SPL_)X86_16BIT_INIT) += arch/x86/cpu/start16.o
-head-$(CONFIG_$(SPL_)X86_16BIT_INIT) += arch/x86/cpu/resetvec.o
+head-$(CONFIG_$(SPL_TPL_)X86_16BIT_INIT) += arch/x86/cpu/start16.o
+head-$(CONFIG_$(SPL_TPL_)X86_16BIT_INIT) += arch/x86/cpu/resetvec.o
 
 libs-y += arch/x86/cpu/
 libs-y += arch/x86/lib/
arch/x86/cpu/intel_common/Makefile +1 -1
index bf798c287f3..80fbc7ab661 100644
@@ -3,7 +3,7 @@
 # Copyright (c) 2016 Google, Inc
 
 ifdef CONFIG_HAVE_MRC
-obj-$(CONFIG_$(SPL_)X86_32BIT_INIT) += car.o
+obj-$(CONFIG_$(SPL_TPL_)X86_16BIT_INIT) += car.o
 obj-$(CONFIG_$(SPL_)X86_32BIT_INIT) += me_status.o
 obj-$(CONFIG_$(SPL_)X86_32BIT_INIT) += report_platform.o
 obj-$(CONFIG_$(SPL_)X86_32BIT_INIT) += mrc.o
arch/x86/cpu/u-boot-spl.lds +1 -1
index 4e656dc4e5a..f20c0b810d3 100644
@@ -54,7 +54,7 @@ SECTIONS
 	/DISCARD/ : { *(.interp*) }
 	/DISCARD/ : { *(.gnu*) }
 
-#ifdef CONFIG_SPL_X86_16BIT_INIT
+#if defined(CONFIG_SPL_X86_16BIT_INIT) || defined(CONFIG_TPL_X86_16BIT_INIT)
 	/*
 	 * The following expressions place the 16-bit Real-Mode code and
 	 * Reset Vector at the end of the Flash ROM
scripts/Makefile.spl +9 -1
index 8d76a2d4007..7af6b120b6c 100644
@@ -228,7 +228,11 @@ ifeq ($(CONFIG_SYS_SOC),"at91")
 ALL-y	+= $(obj)/boot.bin
 endif
 
+ifdef CONFIG_TPL_BUILD
+ALL-$(CONFIG_TPL_X86_16BIT_INIT) += $(obj)/u-boot-x86-16bit-tpl.bin
+else
 ALL-$(CONFIG_SPL_X86_16BIT_INIT) += $(obj)/u-boot-x86-16bit-spl.bin
+endif
 
 ALL-$(CONFIG_ARCH_ZYNQ)		+= $(obj)/boot.bin
 ALL-$(CONFIG_ARCH_ZYNQMP)	+= $(obj)/boot.bin
@@ -328,7 +332,7 @@ quiet_cmd_objcopy = OBJCOPY $@
 cmd_objcopy = $(OBJCOPY) $(OBJCOPYFLAGS) $(OBJCOPYFLAGS_$(@F)) $< $@
 
 OBJCOPYFLAGS_$(SPL_BIN)-nodtb.bin = $(SPL_OBJCFLAGS) -O binary \
-		$(if $(CONFIG_SPL_X86_16BIT_INIT),-R .start16 -R .resetvec)
+		$(if $(CONFIG_$(SPL_TPL_)X86_16BIT_INIT),-R .start16 -R .resetvec)
 
 $(obj)/$(SPL_BIN)-nodtb.bin: $(obj)/$(SPL_BIN) FORCE
 	$(call if_changed,objcopy)
@@ -337,6 +341,10 @@ OBJCOPYFLAGS_u-boot-x86-16bit-spl.bin := -O binary -j .start16 -j .resetvec
 $(obj)/u-boot-x86-16bit-spl.bin: $(obj)/u-boot-spl FORCE
 	$(call if_changed,objcopy)
 
+OBJCOPYFLAGS_u-boot-x86-16bit-tpl.bin := -O binary -j .start16 -j .resetvec
+$(obj)/u-boot-x86-16bit-tpl.bin: $(obj)/u-boot-tpl FORCE
+	$(call if_changed,objcopy)
+
 LDFLAGS_$(SPL_BIN) += -T u-boot-spl.lds $(LDFLAGS_FINAL)
 
 # Avoid 'Not enough room for program headers' error on binutils 2.28 onwards.
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#48 lore raw

PATCH v2 21/50 x86: Allow 16-bit init to be in TPL

Quoted 8 lines · Simon Glass

At present we support having 16-bit init be in SPL or U-Boot proper, but not TPL. Add support for this so that TPL can boot.

---

Quoted 9 lines

Changes in v2: None

 Makefile                           |  1 +
 arch/x86/Makefile                  |  4 ++--
 arch/x86/cpu/intel_common/Makefile |  2 +-
 arch/x86/cpu/u-boot-spl.lds        |  2 +-
 scripts/Makefile.spl               | 10 +++++++++-
 5 files changed, 14 insertions(+), 5 deletions(-)

applied to u-boot-x86, thanks!

#49 lore raw

PATCH v2 22/50 x86: broadwell: Allow booting from SPL

In reply to Simon Glass

At present broadwell only supports booting straight into U-Boot proper. Add a separate init file to boot from SPL into U-Boot proper, and select it when SPL is in use.

---

Changes in v2: None

 arch/x86/cpu/broadwell/Makefile       | 15 +++++++
 arch/x86/cpu/broadwell/cpu_from_spl.c | 63 +++++++++++++++++++++++++++
 2 files changed, 78 insertions(+)
 create mode 100644 arch/x86/cpu/broadwell/cpu_from_spl.c
arch/x86/cpu/broadwell/Makefile +15 -0
index 8553615f67b..fbc778cda5f 100644
@@ -5,6 +5,21 @@
 obj-y += adsp.o
 obj-y += cpu.o
 obj-y += cpu_full.o
+
+ifdef CONFIG_SPL
+ifndef CONFIG_SPL_BUILD
+obj-y += cpu_from_spl.o
+obj-y += cpu_full.o
+endif
+ifndef CONFIG_SPL_BUILD
+# obj-y += cpu_from_spl.o
+endif
+endif
+
+ifeq ($(CONFIG_$(SPL_TPL_)X86_32BIT_INIT),)
+#obj-y += cpu_from_spl.o
+endif
+
 obj-y += iobp.o
 obj-y += lpc.o
 obj-y += me.o
arch/x86/cpu/broadwell/cpu_from_spl.c new file +63 -0

63 lines not on this page · raw message

Signature · 1 line
2.21.0.593.g511ec345e18-goog
#50 lore raw

PATCH v2 22/50 x86: broadwell: Allow booting from SPL

Quoted 7 lines · Simon Glass

At present broadwell only supports booting straight into U-Boot proper. Add a separate init file to boot from SPL into U-Boot proper, and select it when SPL is in use.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/cpu/broadwell/Makefile       | 15 +++++++
 arch/x86/cpu/broadwell/cpu_from_spl.c | 63 +++++++++++++++++++++++++++
 2 files changed, 78 insertions(+)
 create mode 100644 arch/x86/cpu/broadwell/cpu_from_spl.c

applied to u-boot-x86, thanks!

#51 lore raw

PATCH v2 23/50 x86: broadwell: Select refcode and CPU code for SPL

In reply to Simon Glass

Allow broadwell to build for SPL and include the reference code.

---

Changes in v2: None

 arch/x86/cpu/broadwell/Makefile | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
arch/x86/cpu/broadwell/Makefile +4 -3
index fbc778cda5f..52d56c65be8 100644
@@ -3,13 +3,14 @@
 # Copyright (c) 2016 Google, Inc
 
 obj-y += adsp.o
-obj-y += cpu.o
-obj-y += cpu_full.o
+obj-$(CONFIG_$(SPL_TPL_)X86_16BIT_INIT) += cpu.o
+obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += cpu_full.o
 
 ifdef CONFIG_SPL
 ifndef CONFIG_SPL_BUILD
 obj-y += cpu_from_spl.o
 obj-y += cpu_full.o
+obj-y += refcode.o
 endif
 ifndef CONFIG_SPL_BUILD
 # obj-y += cpu_from_spl.o
@@ -27,6 +28,6 @@ obj-y += northbridge.o
 obj-y += pch.o
 obj-y += pinctrl_broadwell.o
 obj-y += power_state.o
-obj-y += refcode.o
+obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += refcode.o
 obj-y += sata.o
 obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += sdram.o
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#52 lore raw

PATCH v2 23/50 x86: broadwell: Select refcode and CPU code for SPL

Quoted 12 lines · Simon Glass

Allow broadwell to build for SPL and include the reference code.

---

Changes in v2: None

 arch/x86/cpu/broadwell/Makefile | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

applied to u-boot-x86, thanks!

#53 lore raw

PATCH v2 24/50 x86: Add common Intel code for SPL

In reply to Simon Glass

Add an implementation of arch_cpu_init_f() so that the x86 SPL code builds and identifies the CPU.

---

Changes in v2: None

 arch/x86/cpu/intel_common/Makefile       |  6 ++++++
 arch/x86/cpu/intel_common/cpu_from_spl.c | 27 ++++++++++++++++++++++++
 2 files changed, 33 insertions(+)
 create mode 100644 arch/x86/cpu/intel_common/cpu_from_spl.c
arch/x86/cpu/intel_common/Makefile +6 -0
index 80fbc7ab661..57cca0b930c 100644
@@ -14,3 +14,9 @@ ifndef CONFIG_TARGET_EFI_APP
 obj-y += microcode.o
 endif
 obj-y += pch.o
+
+ifdef CONFIG_SPL
+ifndef CONFIG_SPL_BUILD
+obj-y += cpu_from_spl.o
+endif
+endif
arch/x86/cpu/intel_common/cpu_from_spl.c new file +27 -0
new file mode 100644
index 00000000000..a6233c75ce2
@@ -0,0 +1,27 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2016 Google, Inc
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <errno.h>
+#include <asm/cpu_common.h>
+#include <asm/intel_regs.h>
+#include <asm/lapic.h>
+#include <asm/lpc_common.h>
+#include <asm/msr.h>
+#include <asm/mtrr.h>
+#include <asm/post.h>
+#include <asm/microcode.h>
+
+DECLARE_GLOBAL_DATA_PTR;
+
+int arch_cpu_init(void)
+{
+	int ret;
+
+	ret = x86_cpu_reinit_f();
+
+	return ret;
+}
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#54 lore raw

PATCH v2 24/50 x86: Add common Intel code for SPL

Quoted 6 lines · Simon Glass

Add an implementation of arch_cpu_init_f() so that the x86 SPL code builds and identifies the CPU.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/cpu/intel_common/Makefile       |  6 ++++++
 arch/x86/cpu/intel_common/cpu_from_spl.c | 27 ++++++++++++++++++++++++
 2 files changed, 33 insertions(+)
 create mode 100644 arch/x86/cpu/intel_common/cpu_from_spl.c

applied to u-boot-x86, thanks!

#55 lore raw

PATCH v2 25/50 x86: Support saving MRC data from SPL

In reply to Simon Glass

When SPL is used to set up the memory controller we want to save the MRC data in SPL to avoid needing to pass it up to U-Boot proper to save. Add a function to handle that.

---

Changes in v2: None

 arch/x86/include/asm/mrccache.h | 11 ++++++++++
 arch/x86/lib/mrccache.c         | 36 ++++++++++++++++++++++++---------
 2 files changed, 38 insertions(+), 9 deletions(-)
arch/x86/include/asm/mrccache.h +11 -0
index 04783cd3292..40fda856ff4 100644
@@ -103,4 +103,15 @@ int mrccache_get_region(struct udevice **devp, struct mrc_region *entry);
  */
 int mrccache_save(void);
 
+/**
+ * mrccache_spl_save() - Save to the MRC region from SPL
+ *
+ * When SPL is used to set up the memory controller we want to save the MRC
+ * data in SPL to avoid needing to pass it up to U-Boot proper to save. This
+ * function handles that.
+ *
+ * @return 0 if saved to SPI flash successfully, other error if failed
+ */
+int mrccache_spl_save(void);
+
 #endif /* _ASM_MRCCACHE_H */
arch/x86/lib/mrccache.c +27 -9
index 2a8919885b7..f37a732a45c 100644
@@ -159,18 +159,11 @@ int mrccache_update(struct udevice *sf, struct mrc_region *entry,
 	return 0;
 }
 
-int mrccache_reserve(void)
+static void mrccache_setup(void *data)
 {
-	struct mrc_data_container *cache;
+	struct mrc_data_container *cache = data;
 	u16 checksum;
 
-	if (!gd->arch.mrc_output_len)
-		return 0;
-
-	/* adjust stack pointer to store pure cache data plus the header */
-	gd->start_addr_sp -= (gd->arch.mrc_output_len + MRC_DATA_HEADER_SIZE);
-	cache = (struct mrc_data_container *)gd->start_addr_sp;
-
 	cache->signature = MRC_DATA_SIGNATURE;
 	cache->data_size = gd->arch.mrc_output_len;
 	checksum = compute_ip_checksum(gd->arch.mrc_output, cache->data_size);
@@ -182,6 +175,16 @@ int mrccache_reserve(void)
 
 	/* gd->arch.mrc_output now points to the container */
 	gd->arch.mrc_output = (char *)cache;
+}
+
+int mrccache_reserve(void)
+{
+	if (!gd->arch.mrc_output_len)
+		return 0;
+
+	/* adjust stack pointer to store pure cache data plus the header */
+	gd->start_addr_sp -= (gd->arch.mrc_output_len + MRC_DATA_HEADER_SIZE);
+	mrccache_setup((void *)gd->start_addr_sp);
 
 	gd->start_addr_sp &= ~0xf;
 
@@ -256,3 +259,18 @@ err_entry:
 		debug("%s: Failed: %d\n", __func__, ret);
 	return ret;
 }
+
+int mrccache_spl_save(void)
+{
+	void *data;
+	int size;
+
+	size = gd->arch.mrc_output_len + MRC_DATA_HEADER_SIZE;
+	data = malloc(size);
+	if (!data)
+		return log_msg_ret("Allocate MRC cache block", -ENOMEM);
+	mrccache_setup(data);
+	gd->arch.mrc_output = data;
+
+	return mrccache_save();
+}
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#56 lore raw

PATCH v2 25/50 x86: Support saving MRC data from SPL

Quoted 7 lines · Simon Glass

When SPL is used to set up the memory controller we want to save the MRC data in SPL to avoid needing to pass it up to U-Boot proper to save. Add a function to handle that.

Quoted 8 lines

---

Changes in v2: None

 arch/x86/include/asm/mrccache.h | 11 ++++++++++
 arch/x86/lib/mrccache.c         | 36 ++++++++++++++++++++++++---------
 2 files changed, 38 insertions(+), 9 deletions(-)

applied to u-boot-x86, thanks!

#57 lore raw

PATCH v2 26/50 x86: Add a simple TPL implementation

In reply to Simon Glass

Add the required CPU code so that TPL builds correctly. Also update the SPL code to deal with being booted from TPL.

---

Changes in v2: None

 arch/x86/include/asm/spl.h        |  17 ++++-
 arch/x86/lib/Makefile             |   9 ++-
 arch/x86/lib/spl.c                |  44 ++++++++++-
 arch/x86/lib/tpl.c                | 118 ++++++++++++++++++++++++++++++
 include/configs/chromebook_link.h |   3 -
 include/configs/qemu-x86.h        |   3 -
 6 files changed, 183 insertions(+), 11 deletions(-)
 create mode 100644 arch/x86/lib/tpl.c
arch/x86/include/asm/spl.h +15 -2
index 8cf59d14e7c..27432b28979 100644
@@ -2,6 +2,19 @@
 /*
  * Copyright (C) 2017 Google, Inc
  * Written by Simon Glass <[email protected]>
- *
- * This file is required for SPL to build, but is empty.
  */
+
+#ifndef __asm_spl_h
+#define __asm_spl_h
+
+#define CONFIG_SPL_BOARD_LOAD_IMAGE
+
+enum {
+	BOOT_DEVICE_SPI		= 10,
+	BOOT_DEVICE_BOARD,
+	BOOT_DEVICE_CROS_VBOOT,
+};
+
+void jump_to_spl(ulong entry);
+
+#endif
arch/x86/lib/Makefile +8 -1
index 56fd680033b..436252dd831 100644
@@ -43,7 +43,14 @@ ifndef CONFIG_SPL_BUILD
 obj-$(CONFIG_CMD_ZBOOT)	+= zimage.o
 endif
 obj-$(CONFIG_HAVE_FSP) += fsp/
-obj-$(CONFIG_SPL_BUILD) += spl.o
+
+ifdef CONFIG_SPL_BUILD
+ifdef CONFIG_TPL_BUILD
+obj-y += tpl.o
+else
+obj-y += spl.o
+endif
+endif
 
 lib-$(CONFIG_USE_PRIVATE_LIBGCC) += div64.o
 
arch/x86/lib/spl.c +42 -2

116 lines not on this page · raw message

arch/x86/lib/tpl.c new file +118 -0

118 lines not on this page · raw message

include/configs/chromebook_link.h +0 -3
index ee546d8f893..f59ed8e28b1 100644
@@ -20,9 +20,6 @@
 
 #define CONFIG_SPL_TEXT_BASE		0xfffd0000
 
-#define BOOT_DEVICE_SPI			10
-
 #define CONFIG_SPL_BOARD_LOAD_IMAGE
-#define BOOT_DEVICE_BOARD		11
 
 #endif	/* __CONFIG_H */
include/configs/qemu-x86.h +0 -3
index 4b9ddd6f258..59206fb24b1 100644
@@ -37,9 +37,6 @@
 
 #define CONFIG_SPL_TEXT_BASE		0xfffd0000
 
-#define BOOT_DEVICE_SPI			10
-
 #define CONFIG_SPL_BOARD_LOAD_IMAGE
-#define BOOT_DEVICE_BOARD		11
 
 #endif	/* __CONFIG_H */
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#58 lore raw

PATCH v2 26/50 x86: Add a simple TPL implementation

Quoted 10 lines · Simon Glass

Add the required CPU code so that TPL builds correctly. Also update the SPL code to deal with being booted from TPL.

---

Changes in v2: None

Quoted 9 lines
 arch/x86/include/asm/spl.h        |  17 ++++-
 arch/x86/lib/Makefile             |   9 ++-
 arch/x86/lib/spl.c                |  44 ++++++++++-
 arch/x86/lib/tpl.c                | 118 ++++++++++++++++++++++++++++++
 include/configs/chromebook_link.h |   3 -
 include/configs/qemu-x86.h        |   3 -
 6 files changed, 183 insertions(+), 11 deletions(-)
 create mode 100644 arch/x86/lib/tpl.c

This patch does not apply cleanly.

Applying: x86: Add a simple TPL implementation
error: patch failed: include/configs/chromebook_link.h:20
error: include/configs/chromebook_link.h: patch does not apply
error: patch failed: include/configs/qemu-x86.h:37
error: include/configs/qemu-x86.h: patch does not apply Patch failed at 0001 x86: Add a simple TPL implementation

Regards,
Bin

#59 lore raw

PATCH v2 27/50 x86: mrccache: Add more debugging

In reply to Simon Glass

When the MRC cache fails to save it is useful to have some debugging info to indicate what when wrong. Add some more debug() calls.

---

Changes in v2: None

 arch/x86/lib/mrccache.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)
arch/x86/lib/mrccache.c +12 -4
index f37a732a45c..be107627b80 100644
@@ -113,8 +113,10 @@ int mrccache_update(struct udevice *sf, struct mrc_region *entry,
 	ulong base_addr;
 	int ret;
 
-	if (!is_mrc_cache(cur))
+	if (!is_mrc_cache(cur)) {
+		debug("%s: Cache data not valid\n", __func__);
 		return -EINVAL;
+	}
 
 	/* Find the last used block */
 	base_addr = entry->base + entry->offset;
@@ -205,17 +207,23 @@ int mrccache_get_region(struct udevice **devp, struct mrc_region *entry)
 		return -ENOENT;
 	}
 
-	if (fdtdec_get_int_array(blob, node, "memory-map", reg, 2))
+	if (fdtdec_get_int_array(blob, node, "memory-map", reg, 2)) {
+		debug("%s: Cannot find memory map\n", __func__);
 		return -EINVAL;
+	}
 	entry->base = reg[0];
 
 	/* Find the place where we put the MRC cache */
 	mrc_node = fdt_subnode_offset(blob, node, "rw-mrc-cache");
-	if (mrc_node < 0)
+	if (mrc_node < 0) {
+		debug("%s: Cannot find node\n", __func__);
 		return -EPERM;
+	}
 
-	if (fdtdec_get_int_array(blob, mrc_node, "reg", reg, 2))
+	if (fdtdec_get_int_array(blob, mrc_node, "reg", reg, 2)) {
+		debug("%s: Cannot find address\n", __func__);
 		return -EINVAL;
+	}
 	entry->offset = reg[0];
 	entry->length = reg[1];
 
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#60 lore raw

PATCH v2 27/50 x86: mrccache: Add more debugging

Quoted 4 lines · Simon Glass

When the MRC cache fails to save it is useful to have some debugging info to indicate what when wrong. Add some more debug() calls.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/lib/mrccache.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

applied to u-boot-x86, thanks!

#61 lore raw

PATCH v2 28/50 x86: sysreset: Separate out the EFI code

In reply to Simon Glass

The EFI implementation of reset sits inside the driver and is called directly from outside the driver, breaking the normal driver-model conventions. Worse, it passed NULL as the device pointer, hoping that the called function won't use it, which breaks as soon as code is added to use it.

Separate out the implementation to improve the situation enough to allow a future patch to add new sysreset features.

---

Changes in v2:
- Add new patch to separate out the EFI code in sysreset

 drivers/sysreset/sysreset_x86.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)
drivers/sysreset/sysreset_x86.c +11 -5
index 009f3766027..d484ec5de49 100644
@@ -12,8 +12,7 @@
 #include <asm/processor.h>
 #include <efi_loader.h>
 
-static __efi_runtime int x86_sysreset_request(struct udevice *dev,
-					      enum sysreset_t type)
+static int x86_sysreset_request(struct udevice *dev, enum sysreset_t type)
 {
 	int value;
 
@@ -39,11 +38,18 @@ void __efi_runtime EFIAPI efi_reset_system(
 			efi_status_t reset_status,
 			unsigned long data_size, void *reset_data)
 {
+	int value;
+
+	/*
+	 * inline this code since we are not caused in the context of a
+	 * udevice and passing NULL to x86_sysreset_request() is too horrible.
+	 */
 	if (reset_type == EFI_RESET_COLD ||
 		 reset_type == EFI_RESET_PLATFORM_SPECIFIC)
-		x86_sysreset_request(NULL, SYSRESET_COLD);
-	else if (reset_type == EFI_RESET_WARM)
-		x86_sysreset_request(NULL, SYSRESET_WARM);
+		value = SYS_RST | RST_CPU | FULL_RST;
+	else /* assume EFI_RESET_WARM since we cannot return an error */
+		value = SYS_RST | RST_CPU;
+	outb(value, IO_PORT_RESET);
 
 	/* TODO EFI_RESET_SHUTDOWN */
 
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#62 lore raw

PATCH v2 28/50 x86: sysreset: Separate out the EFI code

+Heinrich,

Quoted 24 lines · Simon Glass

The EFI implementation of reset sits inside the driver and is called directly from outside the driver, breaking the normal driver-model conventions. Worse, it passed NULL as the device pointer, hoping that the called function won't use it, which breaks as soon as code is added to use it.

Separate out the implementation to improve the situation enough to allow a future patch to add new sysreset features.

---

Changes in v2:
- Add new patch to separate out the EFI code in sysreset

 drivers/sysreset/sysreset_x86.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)
drivers/sysreset/sysreset_x86.c +0 -0
index 009f3766027..d484ec5de49 100644

drivers/sysreset/sysreset_x86.c line 16

Quoted 5 lines
drivers/sysreset/sysreset_x86.c +0 -2
@@ -12,8 +12,7 @@
 #include <asm/processor.h>
 #include <efi_loader.h>
 
-static __efi_runtime int x86_sysreset_request(struct udevice *dev,
-                                             enum sysreset_t type)

I remember last time I tried when booting Linux from EFI loader on U-Boot, calling from kernel into the EFI runtime was broken. Not sure what the latest status is.

Quoted 17 lines
drivers/sysreset/sysreset_x86.c +7 -1
+static int x86_sysreset_request(struct udevice *dev, enum sysreset_t type)
 {
        int value;
 
@@ -39,11 +38,18 @@ void __efi_runtime EFIAPI efi_reset_system(
                        efi_status_t reset_status,
                        unsigned long data_size, void *reset_data)
 {
+       int value;
+
+       /*
+        * inline this code since we are not caused in the context of a
+        * udevice and passing NULL to x86_sysreset_request() is too horrible.
+        */
        if (reset_type == EFI_RESET_COLD ||
                 reset_type == EFI_RESET_PLATFORM_SPECIFIC)
-               x86_sysreset_request(NULL, SYSRESET_COLD);

drivers/sysreset/sysreset_x86.c void __efi_runtime EFIAPI efi_reset_system(

Quoted 10 lines
drivers/sysreset/sysreset_x86.c +4 -3
-       else if (reset_type == EFI_RESET_WARM)
-               x86_sysreset_request(NULL, SYSRESET_WARM);
+               value = SYS_RST | RST_CPU | FULL_RST;
+       else /* assume EFI_RESET_WARM since we cannot return an error */
+               value = SYS_RST | RST_CPU;
+       outb(value, IO_PORT_RESET);
 
        /* TODO EFI_RESET_SHUTDOWN */
 
--
#63 lore raw

PATCH v2 28/50 x86: sysreset: Separate out the EFI code

Quoted 30 lines · Bin Meng

+Heinrich,

Quoted 27 lines · Simon Glass

The EFI implementation of reset sits inside the driver and is called directly from outside the driver, breaking the normal driver-model conventions. Worse, it passed NULL as the device pointer, hoping that the called function won't use it, which breaks as soon as code is added to use it.

Separate out the implementation to improve the situation enough to allow a future patch to add new sysreset features.

---

Changes in v2:
- Add new patch to separate out the EFI code in sysreset

  drivers/sysreset/sysreset_x86.c | 16 +++++++++++-----
  1 file changed, 11 insertions(+), 5 deletions(-)
drivers/sysreset/sysreset_x86.c +0 -0
index 009f3766027..d484ec5de49 100644
@@ -12,8 +12,7 @@
  #include <asm/processor.h>
  #include <efi_loader.h>
 
Quoted 6 lines

drivers/sysreset/sysreset_x86.c line 16

Quoted 2 lines
drivers/sysreset/sysreset_x86.c +0 -2
-static __efi_runtime int x86_sysreset_request(struct udevice *dev,
-                                             enum sysreset_t type)

I remember last time I tried when booting Linux from EFI loader on U-Boot, calling from kernel into the EFI runtime was broken. Not sure what the latest status is.

I have no x86 system using U-Boot. On ARM I am not aware of problems.

Last year I had no success booting up qemu-x86_defconfig with Linux via EFI. Maybe I should retry.

Best regards

Heinrich

#64 lore raw

PATCH v2 29/50 x86: pch: Add an ioctl to read power-management info

In reply to Simon Glass

Add a new ioctl() request to read information about the power-management system. This can be used to power off the device.

---

Changes in v2:
- Add new patch to add an ioctl to read power-management info

 include/pch.h | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
include/pch.h +18 -0
index 046a5fde3ab..0b44b66df9f 100644
@@ -16,6 +16,9 @@ enum pch_req_t {
 	/* Returns HDA config info if Azalia V1CTL enabled, -ENOENT if not */
 	PCH_REQ_HDA_CONFIG,
 
+	/* Fills out a struct pch_pmbase_info if available */
+	PCH_REQ_PMBASE_INFO,
+
 	PCH_REQ_TEST1,		/* Test requests for sandbox driver */
 	PCH_REQ_TEST2,
 	PCH_REQ_TEST3,
@@ -23,6 +26,21 @@ enum pch_req_t {
 	PCH_REQ_COUNT,		/* Number of ioctrls supported */
 };
 
+/**
+ * struct pch_pmbase_info - Information filled in by PCH_REQ_PMBASE_INFO
+ *
+ * @pmbase: IO address of power-management controller
+ * @gpio0_en_ofs: Offset of GPIO0 enable register
+ * @pm1_sts_ofs: Offset of status register
+ * @pm1_cnt_ofs: Offset of control register
+ */
+struct pch_pmbase_info {
+	u16 base;
+	u8 gpio0_en_ofs;
+	u8 pm1_sts_ofs;
+	u8 pm1_cnt_ofs;
+};
+
 /**
  * struct pch_ops - Operations for the Platform Controller Hub
  *
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#65 lore raw

PATCH v2 29/50 x86: pch: Add an ioctl to read power-management info

Quoted 4 lines · Simon Glass

Add a new ioctl() request to read information about the power-management system. This can be used to power off the device.

Quoted 9 lines

---

Changes in v2:
- Add new patch to add an ioctl to read power-management info

 include/pch.h | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
#66 lore raw

PATCH v2 29/50 x86: pch: Add an ioctl to read power-management info

Quoted 9 lines · Bin Meng
Quoted 7 lines · Simon Glass

Add a new ioctl() request to read information about the power-management system. This can be used to power off the device.

---

Quoted 8 lines
Quoted 6 lines

Changes in v2:
- Add new patch to add an ioctl to read power-management info

 include/pch.h | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

applied to u-boot-x86, thanks!

#67 lore raw

PATCH v2 30/50 x86: ivybridge: Implement PCH_REQ_PMBASE_INFO

In reply to Simon Glass

Implement this ioctl() to support power off.

---

Changes in v2:
- Add new patch to implement PCH_REQ_PMBASE_INFO on ivybridge

 arch/x86/cpu/ivybridge/bd82x6x.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
arch/x86/cpu/ivybridge/bd82x6x.c +15 -0
index ed9bce64168..1cb6cecda9e 100644
@@ -229,6 +229,21 @@ static int bd82x6x_ioctl(struct udevice *dev, enum pch_req_t req, void *data,
 			return -ENOENT;
 
 		return val & RCBA_AUDIO_CONFIG_MASK;
+	case PCH_REQ_PMBASE_INFO: {
+		struct pch_pmbase_info *pm = data;
+		int ret;
+
+		/* Find the base address of the powermanagement registers */
+		ret = dm_pci_read_config16(dev, 0x40, &pm->base);
+		if (ret)
+			return ret;
+		pm->base &= 0xfffe;
+		pm->gpio0_en_ofs = GPE0_EN;
+		pm->pm1_sts_ofs = PM1_STS;
+		pm->pm1_cnt_ofs = PM1_CNT;
+
+		return 0;
+	}
 	default:
 		return -ENOSYS;
 	}
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#68 lore raw

PATCH v2 30/50 x86: ivybridge: Implement PCH_REQ_PMBASE_INFO

Quoted 12 lines · Simon Glass

Implement this ioctl() to support power off.

---

Changes in v2:
- Add new patch to implement PCH_REQ_PMBASE_INFO on ivybridge

 arch/x86/cpu/ivybridge/bd82x6x.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
#69 lore raw

PATCH v2 30/50 x86: ivybridge: Implement PCH_REQ_PMBASE_INFO

Quoted 8 lines · Bin Meng
Quoted 6 lines · Simon Glass

Implement this ioctl() to support power off.

---

Quoted 8 lines
Quoted 6 lines

Changes in v2:
- Add new patch to implement PCH_REQ_PMBASE_INFO on ivybridge

 arch/x86/cpu/ivybridge/bd82x6x.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

applied to u-boot-x86, thanks!

#70 lore raw

PATCH v2 31/50 x86: broadwell: Implement PCH_REQ_PMBASE_INFO

In reply to Simon Glass

Implement this ioctl() to support power off.

---

Changes in v2:
- Add new patch to implement PCH_REQ_PMBASE_INFO on broadwell

 arch/x86/cpu/broadwell/pch.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)
arch/x86/cpu/broadwell/pch.c +25 -0
index 73d3d3b5154..e61efa7b16c 100644
@@ -630,10 +630,35 @@ static int broadwell_get_gpio_base(struct udevice *dev, u32 *gbasep)
 	return 0;
 }
 
+static int broadwell_ioctl(struct udevice *dev, enum pch_req_t req, void *data,
+			   int size)
+{
+	switch (req) {
+	case PCH_REQ_PMBASE_INFO: {
+		struct pch_pmbase_info *pm = data;
+		int ret;
+
+		/* Find the base address of the powermanagement registers */
+		ret = dm_pci_read_config16(dev, 0x40, &pm->base);
+		if (ret)
+			return ret;
+		pm->base &= 0xfffe;
+		pm->gpio0_en_ofs = GPE0_EN(0);
+		pm->pm1_sts_ofs = PM1_STS;
+		pm->pm1_cnt_ofs = PM1_CNT;
+
+		return 0;
+	}
+	default:
+		return -ENOSYS;
+	}
+}
+
 static const struct pch_ops broadwell_pch_ops = {
 	.get_spi_base	= broadwell_pch_get_spi_base,
 	.set_spi_protect = broadwell_set_spi_protect,
 	.get_gpio_base	= broadwell_get_gpio_base,
+	.ioctl		= broadwell_ioctl,
 };
 
 static const struct udevice_id broadwell_pch_ids[] = {
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#71 lore raw

PATCH v2 31/50 x86: broadwell: Implement PCH_REQ_PMBASE_INFO

Quoted 12 lines · Simon Glass

Implement this ioctl() to support power off.

---

Changes in v2:
- Add new patch to implement PCH_REQ_PMBASE_INFO on broadwell

 arch/x86/cpu/broadwell/pch.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)
#72 lore raw

PATCH v2 31/50 x86: broadwell: Implement PCH_REQ_PMBASE_INFO

Quoted 8 lines · Bin Meng
Quoted 6 lines · Simon Glass

Implement this ioctl() to support power off.

---

Quoted 8 lines
Quoted 6 lines

Changes in v2:
- Add new patch to implement PCH_REQ_PMBASE_INFO on broadwell

 arch/x86/cpu/broadwell/pch.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)

applied to u-boot-x86, thanks!

#73 lore raw

PATCH v2 32/50 x86: sysreset: Implement power-off if available

In reply to Simon Glass

On modern x86 devices we can power the system off using the power- management features of the PCH. Add an implementation for this.

---

Changes in v2:
- Add new patch to implement power-off if available

 drivers/sysreset/sysreset_x86.c | 82 +++++++++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)
drivers/sysreset/sysreset_x86.c +82 -0
index d484ec5de49..bd759aa8bf4 100644
@@ -7,14 +7,80 @@
 
 #include <common.h>
 #include <dm.h>
+#include <pch.h>
 #include <sysreset.h>
 #include <asm/io.h>
 #include <asm/processor.h>
 #include <efi_loader.h>
 
+struct x86_sysreset_platdata {
+	struct udevice *pch;
+};
+
+#define   PWRBTN_STS	(1 << 8)
+
+#define   SLP_EN	(1 << 13)
+#define   SLP_TYP	(7 << 10)
+#define    SLP_TYP_S5	7
+
+/*
+ * Power down the machine by using the power management sleep control
+ * of the chipset. This will currently only work on Intel chipsets.
+ * However, adapting it to new chipsets is fairly simple. You will
+ * have to find the IO address of the power management register block
+ * in your southbridge, and look up the appropriate SLP_TYP_S5 value
+ * from your southbridge's data sheet.
+ *
+ * This function never returns.
+ */
+int pch_sysreset_power_off(struct udevice *dev)
+{
+	struct x86_sysreset_platdata *plat = dev_get_platdata(dev);
+	struct pch_pmbase_info pm;
+	u32 reg32;
+	int ret;
+
+	if (!plat->pch)
+		return -ENOENT;
+	ret = pch_ioctl(plat->pch, PCH_REQ_PMBASE_INFO, &pm, sizeof(pm));
+	if (ret)
+		return ret;
+
+
+	/* Mask interrupts or system might stay in a coma
+	 * (not executing code anymore, but not powered off either)
+	 */
+	asm("cli");
+
+	/*
+	 * Avoid any GPI waking the system from S5* or the system might stay in
+	 * a coma
+	 */
+	outl(0x00000000, pm.base + pm.gpio0_en_ofs);
+
+	/* Clear Power Button Status */
+	outw(PWRBTN_STS, pm.base + pm.pm1_sts_ofs);
+
+	/* PMBASE + 4, Bit 10-12, Sleeping Type, * set to 111 -> S5, soft_off */
+	reg32 = inl(pm.base + pm.pm1_cnt_ofs);
+
+	/* Set Sleeping Type to S5 (poweroff) */
+	reg32 &= ~(SLP_EN | SLP_TYP);
+	reg32 |= SLP_TYP_S5;
+	outl(reg32, pm.base + pm.pm1_cnt_ofs);
+
+	/* Now set the Sleep Enable bit */
+	reg32 |= SLP_EN;
+	outl(reg32, pm.base + pm.pm1_cnt_ofs);
+
+	for (;;)
+		asm("hlt");
+}
+
 static int x86_sysreset_request(struct udevice *dev, enum sysreset_t type)
 {
 	int value;
+	int ret;
 
 	switch (type) {
 	case SYSRESET_WARM:
@@ -23,6 +89,11 @@ static int x86_sysreset_request(struct udevice *dev, enum sysreset_t type)
 	case SYSRESET_COLD:
 		value = SYS_RST | RST_CPU | FULL_RST;
 		break;
+	case SYSRESET_POWER_OFF:
+		ret = pch_sysreset_power_off(dev);
+		if (ret)
+			return ret;
+		return -EINPROGRESS;
 	default:
 		return -ENOSYS;
 	}
@@ -57,6 +128,15 @@ void __efi_runtime EFIAPI efi_reset_system(
 }
 #endif
 
+static int x86_sysreset_probe(struct udevice *dev)
+{
+	struct x86_sysreset_platdata *plat = dev_get_platdata(dev);
+
+	/* Locate the PCH if there is one. It isn't essential */
+	uclass_first_device(UCLASS_PCH, &plat->pch);
+
+	return 0;
+}
 
 static const struct udevice_id x86_sysreset_ids[] = {
 	{ .compatible = "x86,reset" },
@@ -72,4 +152,6 @@ U_BOOT_DRIVER(x86_sysreset) = {
 	.id = UCLASS_SYSRESET,
 	.of_match = x86_sysreset_ids,
 	.ops = &x86_sysreset_ops,
+	.probe = x86_sysreset_probe,
+	.platdata_auto_alloc_size	= sizeof(struct x86_sysreset_platdata),
 };
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#74 lore raw

PATCH v2 32/50 x86: sysreset: Implement power-off if available

Hi Simon,

Quoted 27 lines · Simon Glass

On modern x86 devices we can power the system off using the power- management features of the PCH. Add an implementation for this.

---

Changes in v2:
- Add new patch to implement power-off if available

 drivers/sysreset/sysreset_x86.c | 82 +++++++++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)
drivers/sysreset/sysreset_x86.c +1 -0
index d484ec5de49..bd759aa8bf4 100644
@@ -7,14 +7,80 @@
 
 #include <common.h>
 #include <dm.h>
+#include <pch.h>
 #include <sysreset.h>
 #include <asm/io.h>
 #include <asm/processor.h>
 #include <efi_loader.h>
 

drivers/sysreset/sysreset_x86.c line 25

Quoted 10 lines
drivers/sysreset/sysreset_x86.c +10 -0
+struct x86_sysreset_platdata {
+       struct udevice *pch;
+};
+
+#define   PWRBTN_STS   (1 << 8)
+
+#define   SLP_EN       (1 << 13)
+#define   SLP_TYP      (7 << 10)
+#define    SLP_TYP_S5  7
+

There are macros from acpi_s3.h that can be used directly here.

Quoted 15 lines
drivers/sysreset/sysreset_x86.c +15 -0
+/*
+ * Power down the machine by using the power management sleep control
+ * of the chipset. This will currently only work on Intel chipsets.
+ * However, adapting it to new chipsets is fairly simple. You will
+ * have to find the IO address of the power management register block
+ * in your southbridge, and look up the appropriate SLP_TYP_S5 value
+ * from your southbridge's data sheet.
+ *
+ * This function never returns.
+ */
+int pch_sysreset_power_off(struct udevice *dev)
+{
+       struct x86_sysreset_platdata *plat = dev_get_platdata(dev);
+       struct pch_pmbase_info pm;
+       u32 reg32;

drivers/sysreset/sysreset_x86.c line 50

Quoted 10 lines
drivers/sysreset/sysreset_x86.c +10 -0
+       int ret;
+
+       if (!plat->pch)
+               return -ENOENT;
+       ret = pch_ioctl(plat->pch, PCH_REQ_PMBASE_INFO, &pm, sizeof(pm));
+       if (ret)
+               return ret;
+
+
+       /* Mask interrupts or system might stay in a coma

nits: wrong multi-line comment format

Quoted 65 lines
drivers/sysreset/sysreset_x86.c +44 -0
+        * (not executing code anymore, but not powered off either)
+        */
+       asm("cli");
+
+       /*
+        * Avoid any GPI waking the system from S5* or the system might stay in
+        * a coma
+        */
+       outl(0x00000000, pm.base + pm.gpio0_en_ofs);
+
+       /* Clear Power Button Status */
+       outw(PWRBTN_STS, pm.base + pm.pm1_sts_ofs);
+
+       /* PMBASE + 4, Bit 10-12, Sleeping Type, * set to 111 -> S5, soft_off */
+       reg32 = inl(pm.base + pm.pm1_cnt_ofs);
+
+       /* Set Sleeping Type to S5 (poweroff) */
+       reg32 &= ~(SLP_EN | SLP_TYP);
+       reg32 |= SLP_TYP_S5;
+       outl(reg32, pm.base + pm.pm1_cnt_ofs);
+
+       /* Now set the Sleep Enable bit */
+       reg32 |= SLP_EN;
+       outl(reg32, pm.base + pm.pm1_cnt_ofs);
+
+       for (;;)
+               asm("hlt");
+}
+
 static int x86_sysreset_request(struct udevice *dev, enum sysreset_t type)
 {
        int value;
+       int ret;
 
        switch (type) {
        case SYSRESET_WARM:
@@ -23,6 +89,11 @@ static int x86_sysreset_request(struct udevice *dev, enum sysreset_t type)
        case SYSRESET_COLD:
                value = SYS_RST | RST_CPU | FULL_RST;
                break;
+       case SYSRESET_POWER_OFF:
+               ret = pch_sysreset_power_off(dev);
+               if (ret)
+                       return ret;
+               return -EINPROGRESS;
        default:
                return -ENOSYS;
        }
@@ -57,6 +128,15 @@ void __efi_runtime EFIAPI efi_reset_system(
 }
 #endif
 
+static int x86_sysreset_probe(struct udevice *dev)
+{
+       struct x86_sysreset_platdata *plat = dev_get_platdata(dev);
+
+       /* Locate the PCH if there is one. It isn't essential */
+       uclass_first_device(UCLASS_PCH, &plat->pch);
+
+       return 0;
+}
 
 static const struct udevice_id x86_sysreset_ids[] = {
        { .compatible = "x86,reset" },

drivers/sysreset/sysreset_x86.c U_BOOT_DRIVER(x86_sysreset) = {

Quoted 7 lines
drivers/sysreset/sysreset_x86.c +2 -1
@@ -72,4 +152,6 @@ U_BOOT_DRIVER(x86_sysreset) = {
        .id = UCLASS_SYSRESET,
        .of_match = x86_sysreset_ids,
        .ops = &x86_sysreset_ops,
+       .probe = x86_sysreset_probe,
+       .platdata_auto_alloc_size       = sizeof(struct x86_sysreset_platdata),
 };
--

Other than above two,

Regards,
Bin

#75 lore raw

PATCH v2 33/50 x86: Support TPL in Intel common code

In reply to Simon Glass

Update the Makefie rules to ensure that the correct files are built when TPL is being used.

---

Changes in v2: None

 arch/x86/cpu/intel_common/Makefile | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
arch/x86/cpu/intel_common/Makefile +6 -3
index 57cca0b930c..07f27c29ec7 100644
@@ -4,15 +4,18 @@
 
 ifdef CONFIG_HAVE_MRC
 obj-$(CONFIG_$(SPL_TPL_)X86_16BIT_INIT) += car.o
-obj-$(CONFIG_$(SPL_)X86_32BIT_INIT) += me_status.o
-obj-$(CONFIG_$(SPL_)X86_32BIT_INIT) += report_platform.o
-obj-$(CONFIG_$(SPL_)X86_32BIT_INIT) += mrc.o
+obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += me_status.o
+obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += report_platform.o
+obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += mrc.o
 endif
 obj-y += cpu.o
 obj-y += lpc.o
 ifndef CONFIG_TARGET_EFI_APP
+obj-$(CONFIG_$(SPL_TPL_)X86_32BIT_INIT) += microcode.o
+ifndef CONFIG_$(SPL_)X86_64
 obj-y += microcode.o
 endif
+endif
 obj-y += pch.o
 
 ifdef CONFIG_SPL
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#76 lore raw

PATCH v2 33/50 x86: Support TPL in Intel common code

Quoted 4 lines · Simon Glass

Update the Makefie rules to ensure that the correct files are built when TPL is being used.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/cpu/intel_common/Makefile | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

applied to u-boot-x86, thanks!

#77 lore raw

PATCH v2 34/50 x86: Don't set up MTRRs in SPL

In reply to Simon Glass

The MTRRs are normally set up in U-Boot proper, so avoid setting them up in SPL as well.

---

Changes in v2: None

 arch/x86/lib/init_helpers.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
arch/x86/lib/init_helpers.c +4 -1
index 0481f453ca6..ac85278cdf0 100644
@@ -18,7 +18,10 @@ __weak ulong board_get_usable_ram_top(ulong total_size)
 
 int init_cache_f_r(void)
 {
-#if CONFIG_IS_ENABLED(X86_32BIT_INIT) && !defined(CONFIG_HAVE_FSP)
+#if (CONFIG_IS_ENABLED(X86_32BIT_INIT) || \
+     (!defined(CONFIG_SPL_BUILD) && \
+      !CONFIG_IS_ENABLED(CONFIG_X86_RUN_64BIT))) && \
+    !defined(CONFIG_HAVE_FSP)
 	int ret;
 
 	ret = mtrr_commit(false);
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#78 lore raw

PATCH v2 34/50 x86: Don't set up MTRRs in SPL

Quoted 4 lines · Simon Glass

The MTRRs are normally set up in U-Boot proper, so avoid setting them up in SPL as well.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/lib/init_helpers.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

applied to u-boot-x86, thanks!

#79 lore raw

PATCH v2 35/50 x86: Don't generate a bootstage report in SPL

In reply to Simon Glass

This report is normally generated by U-Boot proper. Correct the condition here so that it respects the Kconfig options for bootstage.

---

Changes in v2: None

 arch/x86/lib/bootm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/lib/bootm.c +1 -1
index 832b1f901c6..5443a862ab5 100644
@@ -35,7 +35,7 @@ void bootm_announce_and_cleanup(void)
 	timestamp_add_now(TS_U_BOOT_START_KERNEL);
 #endif
 	bootstage_mark_name(BOOTSTAGE_ID_BOOTM_HANDOFF, "start_kernel");
-#ifdef CONFIG_BOOTSTAGE_REPORT
+#if CONFIG_IS_ENABLED(BOOTSTAGE_REPORT)
 	bootstage_report();
 #endif
 
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#80 lore raw

PATCH v2 35/50 x86: Don't generate a bootstage report in SPL

Quoted 4 lines · Simon Glass

This report is normally generated by U-Boot proper. Correct the condition here so that it respects the Kconfig options for bootstage.

Quoted 9 lines

---

Changes in v2: None

 arch/x86/lib/bootm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

applied to u-boot-x86, thanks!

#81 lore raw

PATCH v2 36/50 x86: Support PCI VGA ROM when TPL is used

In reply to Simon Glass

When TPL is in use, U-Boot proper should support initing the VGA ROM even though the 32-bit init portion is in SPL. Update the condition to handle this.

---

Changes in v2: None

 drivers/pci/pci_rom.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/pci/pci_rom.c +1 -1
index 7d9b75c2c45..2cede1211bb 100644
@@ -306,7 +306,7 @@ int dm_pci_run_vga_bios(struct udevice *dev, int (*int15_handler)(void),
 			goto err;
 #endif
 	} else {
-#if defined(CONFIG_X86) && CONFIG_IS_ENABLED(X86_32BIT_INIT)
+#if defined(CONFIG_X86) && (CONFIG_IS_ENABLED(X86_32BIT_INIT) || CONFIG_TPL)
 		bios_set_interrupt_handler(0x15, int15_handler);
 
 		bios_run_on_x86(dev, (unsigned long)ram, vesa_mode,
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#82 lore raw

PATCH v2 36/50 x86: Support PCI VGA ROM when TPL is used

Quoted 5 lines · Simon Glass

When TPL is in use, U-Boot proper should support initing the VGA ROM even though the 32-bit init portion is in SPL. Update the condition to handle this.

Quoted 9 lines

---

Changes in v2: None

 drivers/pci/pci_rom.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

applied to u-boot-x86, thanks!

#83 lore raw

PATCH v2 37/50 x86: sysreset: Implement the get_last() method

In reply to Simon Glass

Add a default implementation of this method which always indicates that the last reset was a power-on reset. This is the most likely type of reset and without a PCH-specific driver we cannot determine any other type.

---

Changes in v2: None

 drivers/sysreset/sysreset_x86.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/sysreset/sysreset_x86.c +6 -0
index bd759aa8bf4..b8876dc51e7 100644
@@ -103,6 +103,11 @@ static int x86_sysreset_request(struct udevice *dev, enum sysreset_t type)
 	return -EINPROGRESS;
 }
 
+static int x86_sysreset_get_last(struct udevice *dev)
+{
+	return SYSRESET_POWER;
+}
+
 #ifdef CONFIG_EFI_LOADER
 void __efi_runtime EFIAPI efi_reset_system(
 			enum efi_reset_type reset_type,
@@ -145,6 +150,7 @@ static const struct udevice_id x86_sysreset_ids[] = {
 
 static struct sysreset_ops x86_sysreset_ops = {
 	.request = x86_sysreset_request,
+	.get_last = x86_sysreset_get_last,
 };
 
 U_BOOT_DRIVER(x86_sysreset) = {
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#84 lore raw

PATCH v2 37/50 x86: sysreset: Implement the get_last() method

Quoted 5 lines · Simon Glass

Add a default implementation of this method which always indicates that the last reset was a power-on reset. This is the most likely type of reset and without a PCH-specific driver we cannot determine any other type.

Quoted 8 lines

---

Changes in v2: None

 drivers/sysreset/sysreset_x86.c | 6 ++++++
 1 file changed, 6 insertions(+)
#85 lore raw

PATCH v2 38/50 x86: Add documention on the samus flashmap

In reply to Simon Glass

There are quite a few variables which control where things appear in the final ROM image. Add a flashmap in the documentation to make this easier to figure out.

---

Changes in v2: None

 doc/README.x86 | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
doc/README.x86 +14 -0
index fa49cb8b8a2..d5224b75367 100644
@@ -185,6 +185,20 @@ If you are using em100, then this command will flash write -Boot:
 
    em100 -s -d filename.rom -c W25Q64CV -r
 
+Flash map for samus / broadwell:
+
+   fffff800	SYS_X86_START16
+   ffff0000	RESET_SEG_START
+   fffd8000	TPL_TEXT_BASE
+   fffa0000	X86_MRC_ADDR
+   fff90000	VGA_BIOS_ADDR
+   ffed0000	SYS_TEXT_BASE
+   ffea0000	X86_REFCODE_ADDR
+   ffe70000	SPL_TEXT_BASE
+   ffa00000	<spare>
+   ff801000	intel-me (address set by descriptor.bin)
+   ff800000	intel-descriptor
+
 ---
 
 Intel Crown Bay specific instructions for bare mode:
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#86 lore raw

PATCH v2 39/50 x86: samus: Update device tree for SPL

In reply to Simon Glass

Add tags to allow required nodes to be present in SPL / TPL. Also enable the sysreset driver.

---

Changes in v2: None

 arch/x86/dts/chromebook_samus.dts | 38 +++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)
arch/x86/dts/chromebook_samus.dts +34 -4
index 35211ed81b1..6664c3b7d36 100644
@@ -17,6 +17,7 @@
 		spi0 = &spi;
 		usb0 = &usb_0;
 		usb1 = &usb_1;
+		cros-ec0 = &cros_ec;
 	};
 
 	config {
@@ -73,6 +74,7 @@
 
 		/* Put this first: it is the default */
 		gpio_unused: gpio-unused {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_INPUT>;
 			owner = <OWNER_GPIO>;
@@ -80,6 +82,7 @@
 		};
 
 		gpio_acpi_sci: acpi-sci {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_INPUT>;
 			invert;
@@ -87,6 +90,7 @@
 		};
 
 		gpio_acpi_smi: acpi-smi {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_INPUT>;
 			invert;
@@ -94,12 +98,14 @@
 		};
 
 		gpio_input: gpio-input {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_INPUT>;
 			owner = <OWNER_GPIO>;
 		};
 
 		gpio_input_invert: gpio-input-invert {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_INPUT>;
 			owner = <OWNER_GPIO>;
@@ -107,9 +113,11 @@
 		};
 
 		gpio_native: gpio-native {
+			u-boot,dm-pre-reloc;
 		};
 
 		gpio_out_high: gpio-out-high {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_OUTPUT>;
 			output-value = <1>;
@@ -118,6 +126,7 @@
 		};
 
 		gpio_out_low: gpio-out-low {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_OUTPUT>;
 			output-value = <0>;
@@ -126,6 +135,7 @@
 		};
 
 		gpio_pirq: gpio-pirq {
+			u-boot,dm-pre-reloc;
 			mode-gpio;
 			direction = <PIN_INPUT>;
 			owner = <OWNER_GPIO>;
@@ -133,6 +143,7 @@
 		};
 
 		soc_gpio at 0 {
+			u-boot,dm-pre-reloc;
 			config =
 				<0 &gpio_unused 0>,	/* unused */
 				<1 &gpio_unused 0>,	/* unused */
@@ -250,8 +261,10 @@
 			spd {
 				#address-cells = <1>;
 				#size-cells = <0>;
+				u-boot,dm-pre-reloc;
 				samsung_4 {
 					reg = <6>;
+					u-boot,dm-pre-reloc;
 					data = [91 20 f1 03 04 11 05 0b
 						03 11 01 08 0a 00 50 01
 						78 78 90 50 90 11 50 e0
@@ -291,6 +304,7 @@
 					 * columns 10, density 4096 mb, x32
 					 */
 					reg = <8>;
+					u-boot,dm-pre-reloc;
 					data = [91 20 f1 03 04 11 05 0b
 						03 11 01 08 0a 00 50 01
 						78 78 90 50 90 11 50 e0
@@ -326,6 +340,7 @@
 					};
 				samsung_8 {
 					reg = <10>;
+					u-boot,dm-pre-reloc;
 					data = [91 20 f1 03 04 12 05 0a
 						03 11 01 08 0a 00 50 01
 						78 78 90 50 90 11 50 e0
@@ -365,6 +380,7 @@
 					 * columns 11, density 4096 mb, x16
 					 */
 					reg = <12>;
+					u-boot,dm-pre-reloc;
 					data = [91 20 f1 03 04 12 05 0a
 						03 11 01 08 0a 00 50 01
 						78 78 90 50 90 11 50 e0
@@ -404,6 +420,7 @@
 					 * columns 11, density 8192 mb, x16
 					 */
 					reg = <13>;
+					u-boot,dm-pre-reloc;
 					data = [91 20 f1 03 05 1a 05 0a
 						03 11 01 08 0a 00 50 01
 						78 78 90 50 90 11 50 e0
@@ -443,6 +460,7 @@
 					 * columns 11, density 8192 mb, x16
 					 */
 					reg = <15>;
+					u-boot,dm-pre-reloc;
 					data = [91 20 f1 03 05 1a 05 0a
 						03 11 01 08 0a 00 50 01
 						78 78 90 50 90 11 50 e0
@@ -540,7 +558,7 @@
 			compatible = "ehci-pci";
 		};
 
-		pch at 1f,0 {
+		pch: pch at 1f,0 {
 			reg = <0x0000f800 0 0 0 0>;
 			compatible = "intel,broadwell-pch";
 			u-boot,dm-pre-reloc;
@@ -559,17 +577,20 @@
 			power-enable-gpio = <&gpio_a 23 0>;
 
 			spi: spi {
+				u-boot,dm-pre-reloc;
 				#address-cells = <1>;
 				#size-cells = <0>;
 				compatible = "intel,ich9-spi";
 				spi-flash at 0 {
+					u-boot,dm-pre-reloc;
 					#size-cells = <1>;
 					#address-cells = <1>;
 					reg = <0>;
 					compatible = "winbond,w25q64",
 							"jedec,spi-nor";
-					memory-map = <0xff800000 0x00800000>;
+					memory-map = <0 0xff800000 0 0x00800000>;
 					rw-mrc-cache {
+						u-boot,dm-pre-reloc;
 						label = "rw-mrc-cache";
 						reg = <0x003e0000 0x00010000>;
 					};
@@ -609,7 +630,8 @@
 				#size-cells = <0>;
 				u-boot,dm-pre-reloc;
 				intel,gen-dec = <0x800 0xfc 0x900 0xfc>;
-				cros-ec at 200 {
+				cros_ec: cros-ec {
+					u-boot,dm-pre-reloc;
 					compatible = "google,cros-ec-lpc";
 					reg = <0x204 1 0x200 1 0x880 0x80>;
 
@@ -630,7 +652,7 @@
 		sata at 1f,2 {
 			compatible = "intel,wildcatpoint-ahci";
 			reg = <0x0000fa00 0 0 0 0>;
-			u-boot,dm-pre-reloc;
+			u-boot,dm-pre-proper;
 			intel,sata-mode = "ahci";
 			intel,sata-port-map = <1>;
 			intel,sata-port0-gen3-tx = <0x72>;
@@ -644,13 +666,21 @@
 		};
 	};
 
+	pch-reset {
+		compatible = "intel,pch-reset";
+		intel,pch = <&pch>;
+	};
+
 	tpm {
+		u-boot,dm-pre-reloc;
 		reg = <0xfed40000 0x5000>;
 		compatible = "infineon,slb9635lpc";
 	};
 
 	microcode {
+		u-boot,dm-pre-reloc;
 		update at 0 {
+			u-boot,dm-pre-reloc;
 #include "microcode/mc0306d4_00000018.dtsi"
 		};
 	};
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#87 lore raw

PATCH v2 39/50 x86: samus: Update device tree for SPL

Hi Simon,

Quoted 202 lines · Simon Glass

Add tags to allow required nodes to be present in SPL / TPL. Also enable the sysreset driver.

---

Changes in v2: None

 arch/x86/dts/chromebook_samus.dts | 38 +++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)
arch/x86/dts/chromebook_samus.dts +26 -4
index 35211ed81b1..6664c3b7d36 100644
@@ -17,6 +17,7 @@
                spi0 = &spi;
                usb0 = &usb_0;
                usb1 = &usb_1;
+               cros-ec0 = &cros_ec;
        };
 
        config {
@@ -73,6 +74,7 @@
 
                /* Put this first: it is the default */
                gpio_unused: gpio-unused {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_INPUT>;
                        owner = <OWNER_GPIO>;
@@ -80,6 +82,7 @@
                };
 
                gpio_acpi_sci: acpi-sci {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_INPUT>;
                        invert;
@@ -87,6 +90,7 @@
                };
 
                gpio_acpi_smi: acpi-smi {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_INPUT>;
                        invert;
@@ -94,12 +98,14 @@
                };
 
                gpio_input: gpio-input {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_INPUT>;
                        owner = <OWNER_GPIO>;
                };
 
                gpio_input_invert: gpio-input-invert {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_INPUT>;
                        owner = <OWNER_GPIO>;
@@ -107,9 +113,11 @@
                };
 
                gpio_native: gpio-native {
+                       u-boot,dm-pre-reloc;
                };
 
                gpio_out_high: gpio-out-high {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_OUTPUT>;
                        output-value = <1>;
@@ -118,6 +126,7 @@
                };
 
                gpio_out_low: gpio-out-low {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_OUTPUT>;
                        output-value = <0>;
@@ -126,6 +135,7 @@
                };
 
                gpio_pirq: gpio-pirq {
+                       u-boot,dm-pre-reloc;
                        mode-gpio;
                        direction = <PIN_INPUT>;
                        owner = <OWNER_GPIO>;
@@ -133,6 +143,7 @@
                };
 
                soc_gpio at 0 {
+                       u-boot,dm-pre-reloc;
                        config =
                                <0 &gpio_unused 0>,     /* unused */
                                <1 &gpio_unused 0>,     /* unused */
@@ -250,8 +261,10 @@
                        spd {
                                #address-cells = <1>;
                                #size-cells = <0>;
+                               u-boot,dm-pre-reloc;
                                samsung_4 {
                                        reg = <6>;
+                                       u-boot,dm-pre-reloc;
                                        data = [91 20 f1 03 04 11 05 0b
                                                03 11 01 08 0a 00 50 01
                                                78 78 90 50 90 11 50 e0
@@ -291,6 +304,7 @@
                                         * columns 10, density 4096 mb, x32
                                         */
                                        reg = <8>;
+                                       u-boot,dm-pre-reloc;
                                        data = [91 20 f1 03 04 11 05 0b
                                                03 11 01 08 0a 00 50 01
                                                78 78 90 50 90 11 50 e0
@@ -326,6 +340,7 @@
                                        };
                                samsung_8 {
                                        reg = <10>;
+                                       u-boot,dm-pre-reloc;
                                        data = [91 20 f1 03 04 12 05 0a
                                                03 11 01 08 0a 00 50 01
                                                78 78 90 50 90 11 50 e0
@@ -365,6 +380,7 @@
                                         * columns 11, density 4096 mb, x16
                                         */
                                        reg = <12>;
+                                       u-boot,dm-pre-reloc;
                                        data = [91 20 f1 03 04 12 05 0a
                                                03 11 01 08 0a 00 50 01
                                                78 78 90 50 90 11 50 e0
@@ -404,6 +420,7 @@
                                         * columns 11, density 8192 mb, x16
                                         */
                                        reg = <13>;
+                                       u-boot,dm-pre-reloc;
                                        data = [91 20 f1 03 05 1a 05 0a
                                                03 11 01 08 0a 00 50 01
                                                78 78 90 50 90 11 50 e0
@@ -443,6 +460,7 @@
                                         * columns 11, density 8192 mb, x16
                                         */
                                        reg = <15>;
+                                       u-boot,dm-pre-reloc;
                                        data = [91 20 f1 03 05 1a 05 0a
                                                03 11 01 08 0a 00 50 01
                                                78 78 90 50 90 11 50 e0
@@ -540,7 +558,7 @@
                        compatible = "ehci-pci";
                };
 
-               pch at 1f,0 {
+               pch: pch at 1f,0 {
                        reg = <0x0000f800 0 0 0 0>;
                        compatible = "intel,broadwell-pch";
                        u-boot,dm-pre-reloc;
@@ -559,17 +577,20 @@
                        power-enable-gpio = <&gpio_a 23 0>;
 
                        spi: spi {
+                               u-boot,dm-pre-reloc;
                                #address-cells = <1>;
                                #size-cells = <0>;
                                compatible = "intel,ich9-spi";
                                spi-flash at 0 {
+                                       u-boot,dm-pre-reloc;
                                        #size-cells = <1>;
                                        #address-cells = <1>;
                                        reg = <0>;
                                        compatible = "winbond,w25q64",
                                                        "jedec,spi-nor";
-                                       memory-map = <0xff800000 0x00800000>;
+                                       memory-map = <0 0xff800000 0 0x00800000>;
                                        rw-mrc-cache {
+                                               u-boot,dm-pre-reloc;
                                                label = "rw-mrc-cache";
                                                reg = <0x003e0000 0x00010000>;
                                        };
@@ -609,7 +630,8 @@
                                #size-cells = <0>;
                                u-boot,dm-pre-reloc;
                                intel,gen-dec = <0x800 0xfc 0x900 0xfc>;
-                               cros-ec at 200 {
+                               cros_ec: cros-ec {
+                                       u-boot,dm-pre-reloc;
                                        compatible = "google,cros-ec-lpc";
                                        reg = <0x204 1 0x200 1 0x880 0x80>;
 
@@ -630,7 +652,7 @@
                sata at 1f,2 {
                        compatible = "intel,wildcatpoint-ahci";
                        reg = <0x0000fa00 0 0 0 0>;
-                       u-boot,dm-pre-reloc;
+                       u-boot,dm-pre-proper;
                        intel,sata-mode = "ahci";
                        intel,sata-port-map = <1>;
                        intel,sata-port0-gen3-tx = <0x72>;

arch/x86/dts/chromebook_samus.dts line 672

Quoted 7 lines
arch/x86/dts/chromebook_samus.dts +4 -0
@@ -644,13 +666,21 @@
                };
        };
 
+       pch-reset {
+               compatible = "intel,pch-reset";
+               intel,pch = <&pch>;
+       };

I think you forgot to remove this node in v2.

[snip]

Regards,
Bin

#88 lore raw

PATCH v2 40/50 x86: samus: Update device tree for verified boot

In reply to Simon Glass

Add nvdata drivers for the TPM and RTC as used on samus. These are needed for Chromium OS verified boot on samus.

---

Changes in v2: None

 arch/x86/dts/chromebook_samus.dts | 22 +++++++++++++++++++++-
 1 file changed, 21 insertions(+), 1 deletion(-)
arch/x86/dts/chromebook_samus.dts +21 -1
index 6664c3b7d36..220bdebfc97 100644
@@ -9,6 +9,12 @@
 /include/ "rtc.dtsi"
 /include/ "tsc_timer.dtsi"
 
+#ifdef CONFIG_CHROMEOS
+#include "chromeos-x86.dtsi"
+#include "flashmap-x86-ro.dtsi"
+#include "flashmap-8mb-rw.dtsi"
+#endif
+
 / {
 	model = "Google Samus";
 	compatible = "google,samus", "intel,broadwell";
@@ -581,7 +587,7 @@
 				#address-cells = <1>;
 				#size-cells = <0>;
 				compatible = "intel,ich9-spi";
-				spi-flash at 0 {
+				fwstore_spi: spi-flash at 0 {
 					u-boot,dm-pre-reloc;
 					#size-cells = <1>;
 					#address-cells = <1>;
@@ -675,6 +681,10 @@
 		u-boot,dm-pre-reloc;
 		reg = <0xfed40000 0x5000>;
 		compatible = "infineon,slb9635lpc";
+		secdata {
+			u-boot,dm-pre-reloc;
+			compatible = "google,tpm-secdata";
+		};
 	};
 
 	microcode {
@@ -698,3 +708,13 @@
 	};
 
 };
+
+&rtc {
+	#address-cells = <1>;
+	#size-cells = <0>;
+	nvdata {
+		u-boot,dm-pre-reloc;
+		compatible = "google,cmos-nvdata";
+		reg = <0x26>;
+	};
+};
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#89 lore raw

PATCH v2 41/50 x86: Update device tree for TPL

In reply to Simon Glass

Add TPL binaries to the device x86 binman desciption. When enabled, TPL will start first, doing the 16-bit init, then jump to SPL and finally U-Boot proper.

---

Changes in v2: None

 arch/x86/dts/u-boot.dtsi | 26 ++++++++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)
arch/x86/dts/u-boot.dtsi +24 -2
index 1050236330a..70e9c8f7acd 100644
@@ -22,7 +22,21 @@
 			filename = CONFIG_INTEL_ME_FILE;
 		};
 #endif
-#ifdef CONFIG_SPL
+#ifdef CONFIG_TPL
+		u-boot-spl {
+			offset = <CONFIG_SPL_TEXT_BASE>;
+		};
+		u-boot-spl-dtb {
+		};
+		u-boot-tpl-with-ucode-ptr {
+			offset = <CONFIG_TPL_TEXT_BASE>;
+		};
+		u-boot-tpl-dtb {
+		};
+		u-boot {
+			offset = <CONFIG_SYS_TEXT_BASE>;
+		};
+#elif defined(CONFIG_SPL)
 		u-boot-spl-with-ucode-ptr {
 			offset = <CONFIG_SPL_TEXT_BASE>;
 		};
@@ -31,7 +45,11 @@
 			type = "u-boot-dtb-with-ucode";
 		};
 		u-boot {
+#if CONFIG_SYS_TEXT_BASE == 0x1110000
 			offset = <0xfff00000>;
+#else
+			offset = <CONFIG_SYS_TEXT_BASE>;
+#endif
 		};
 #else
 		u-boot-with-ucode-ptr {
@@ -77,7 +95,11 @@
 			offset = <CONFIG_X86_REFCODE_ADDR>;
 		};
 #endif
-#ifdef CONFIG_SPL
+#ifdef CONFIG_TPL
+		x86-start16-tpl {
+			offset = <CONFIG_SYS_X86_START16>;
+		};
+#elif defined(CONFIG_SPL)
 		x86-start16-spl {
 			offset = <CONFIG_SYS_X86_START16>;
 		};
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#90 lore raw

PATCH v2 41/50 x86: Update device tree for TPL

Hi Simon,

Quoted 23 lines · Simon Glass

Add TPL binaries to the device x86 binman desciption. When enabled, TPL will start first, doing the 16-bit init, then jump to SPL and finally U-Boot proper.

---

Changes in v2: None

 arch/x86/dts/u-boot.dtsi | 26 ++++++++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)
arch/x86/dts/u-boot.dtsi +1 -1
index 1050236330a..70e9c8f7acd 100644
@@ -22,7 +22,21 @@
                        filename = CONFIG_INTEL_ME_FILE;
                };
 #endif
-#ifdef CONFIG_SPL
+#ifdef CONFIG_TPL

arch/x86/dts/u-boot.dtsi line 35

Quoted 10 lines
arch/x86/dts/u-boot.dtsi +10 -0
+               u-boot-spl {
+                       offset = <CONFIG_SPL_TEXT_BASE>;
+               };
+               u-boot-spl-dtb {
+               };
+               u-boot-tpl-with-ucode-ptr {
+                       offset = <CONFIG_TPL_TEXT_BASE>;
+               };
+               u-boot-tpl-dtb {
+               };

nits: since TPL starts before SPL, can we put the TPL* nodes before SPL* nodes?

arch/x86/dts/u-boot.dtsi line 48

Quoted 12 lines
arch/x86/dts/u-boot.dtsi +5 -0
+               u-boot {
+                       offset = <CONFIG_SYS_TEXT_BASE>;
+               };
+#elif defined(CONFIG_SPL)
                u-boot-spl-with-ucode-ptr {
                        offset = <CONFIG_SPL_TEXT_BASE>;
                };
@@ -31,7 +45,11 @@
                        type = "u-boot-dtb-with-ucode";
                };
                u-boot {
+#if CONFIG_SYS_TEXT_BASE == 0x1110000

The magic number comparison does not look good.

Quoted 11 lines
arch/x86/dts/u-boot.dtsi +3 -0
                        offset = <0xfff00000>;
+#else
+                       offset = <CONFIG_SYS_TEXT_BASE>;
+#endif
                };
 #else
                u-boot-with-ucode-ptr {
@@ -77,7 +95,11 @@
                        offset = <CONFIG_X86_REFCODE_ADDR>;
                };
 #endif

arch/x86/dts/u-boot.dtsi

Quoted 10 lines
arch/x86/dts/u-boot.dtsi +5 -2
-#ifdef CONFIG_SPL
+#ifdef CONFIG_TPL
+               x86-start16-tpl {
+                       offset = <CONFIG_SYS_X86_START16>;
+               };
+#elif defined(CONFIG_SPL)
                x86-start16-spl {
                        offset = <CONFIG_SYS_X86_START16>;
                };
--

Regards,
Bin

#91 lore raw

PATCH v2 42/50 x86: Update device tree for Chromium OS verified boot

In reply to Simon Glass

The standard image generated by U-Boot on x86 is u-boot.rom. Add a separate image called image.bin for verified boot. This supports verification in TPL of which SPL/U-Boot to start, then jumping to the correct one, with SPL setting up the SDRAM and U-Boot proper providing the user interface if needed.

---

Changes in v2: None

 arch/x86/dts/u-boot.dtsi | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)
arch/x86/dts/u-boot.dtsi +15 -2
index 70e9c8f7acd..5943619b863 100644
@@ -6,9 +6,23 @@
 
 #include <config.h>
 
-#ifdef CONFIG_ROM_SIZE
+#ifdef CONFIG_CHROMEOS
 / {
 	binman {
+		multiple-images;
+		rom: rom {
+		};
+	};
+};
+#else
+/ {
+	rom: binman {
+	};
+};
+#endif
+
+#ifdef CONFIG_ROM_SIZE
+&rom {
 		filename = "u-boot.rom";
 		end-at-4gb;
 		sort-by-offset;
@@ -108,6 +122,5 @@
 			offset = <CONFIG_SYS_X86_START16>;
 		};
 #endif
-	};
 };
 #endif
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#92 lore raw

PATCH v2 43/50 x86: Fix device-tree indentation

In reply to Simon Glass

With the use of a phandle we can outdent the device tree nodes a little. Fix this.

---

Changes in v2: None

 arch/x86/dts/u-boot.dtsi | 147 +++++++++++++++++++--------------------
 1 file changed, 73 insertions(+), 74 deletions(-)
arch/x86/dts/u-boot.dtsi +73 -74
index 5943619b863..8bb1318a2ce 100644
@@ -23,104 +23,103 @@
 
 #ifdef CONFIG_ROM_SIZE
 &rom {
-		filename = "u-boot.rom";
-		end-at-4gb;
-		sort-by-offset;
-		pad-byte = <0xff>;
-		size = <CONFIG_ROM_SIZE>;
+	filename = "u-boot.rom";
+	end-at-4gb;
+	sort-by-offset;
+	pad-byte = <0xff>;
+	size = <CONFIG_ROM_SIZE>;
 #ifdef CONFIG_HAVE_INTEL_ME
-		intel-descriptor {
-			filename = CONFIG_FLASH_DESCRIPTOR_FILE;
-		};
-		intel-me {
-			filename = CONFIG_INTEL_ME_FILE;
-		};
+	intel-descriptor {
+		filename = CONFIG_FLASH_DESCRIPTOR_FILE;
+	};
+	intel-me {
+		filename = CONFIG_INTEL_ME_FILE;
+	};
 #endif
 #ifdef CONFIG_TPL
-		u-boot-spl {
-			offset = <CONFIG_SPL_TEXT_BASE>;
-		};
-		u-boot-spl-dtb {
-		};
-		u-boot-tpl-with-ucode-ptr {
-			offset = <CONFIG_TPL_TEXT_BASE>;
-		};
-		u-boot-tpl-dtb {
-		};
-		u-boot {
-			offset = <CONFIG_SYS_TEXT_BASE>;
-		};
+	u-boot-spl {
+		offset = <CONFIG_SPL_TEXT_BASE>;
+	};
+	u-boot-spl-dtb {
+	};
+	u-boot-tpl-with-ucode-ptr {
+		offset = <CONFIG_TPL_TEXT_BASE>;
+	};
+	u-boot-tpl-dtb {
+	};
+	u-boot {
+		offset = <CONFIG_SYS_TEXT_BASE>;
+	};
 #elif defined(CONFIG_SPL)
-		u-boot-spl-with-ucode-ptr {
-			offset = <CONFIG_SPL_TEXT_BASE>;
-		};
-
-		u-boot-dtb-with-ucode2 {
-			type = "u-boot-dtb-with-ucode";
-		};
-		u-boot {
+	u-boot-spl-with-ucode-ptr {
+		offset = <CONFIG_SPL_TEXT_BASE>;
+	};
+	u-boot-dtb-with-ucode2 {
+		type = "u-boot-dtb-with-ucode";
+	};
+	u-boot {
 #if CONFIG_SYS_TEXT_BASE == 0x1110000
-			offset = <0xfff00000>;
+		offset = <0xfff00000>;
 #else
-			offset = <CONFIG_SYS_TEXT_BASE>;
+		offset = <CONFIG_SYS_TEXT_BASE>;
 #endif
-		};
+	};
 #else
-		u-boot-with-ucode-ptr {
-			offset = <CONFIG_SYS_TEXT_BASE>;
-		};
+	u-boot-with-ucode-ptr {
+		offset = <CONFIG_SYS_TEXT_BASE>;
+	};
 #endif
-		u-boot-dtb-with-ucode {
-		};
-		u-boot-ucode {
-			align = <16>;
-		};
+	u-boot-dtb-with-ucode {
+	};
+	u-boot-ucode {
+		align = <16>;
+	};
 #ifdef CONFIG_HAVE_MRC
-		intel-mrc {
-			offset = <CONFIG_X86_MRC_ADDR>;
-		};
+	intel-mrc {
+		offset = <CONFIG_X86_MRC_ADDR>;
+	};
 #endif
 #ifdef CONFIG_HAVE_FSP
-		intel-fsp {
-			filename = CONFIG_FSP_FILE;
-			offset = <CONFIG_FSP_ADDR>;
-		};
+	intel-fsp {
+		filename = CONFIG_FSP_FILE;
+		offset = <CONFIG_FSP_ADDR>;
+	};
 #endif
 #ifdef CONFIG_HAVE_CMC
-		intel-cmc {
-			filename = CONFIG_CMC_FILE;
-			offset = <CONFIG_CMC_ADDR>;
-		};
+	intel-cmc {
+		filename = CONFIG_CMC_FILE;
+		offset = <CONFIG_CMC_ADDR>;
+	};
 #endif
 #ifdef CONFIG_HAVE_VGA_BIOS
-		intel-vga {
-			filename = CONFIG_VGA_BIOS_FILE;
-			offset = <CONFIG_VGA_BIOS_ADDR>;
-		};
+	intel-vga {
+		filename = CONFIG_VGA_BIOS_FILE;
+		offset = <CONFIG_VGA_BIOS_ADDR>;
+	};
 #endif
 #ifdef CONFIG_HAVE_VBT
-		intel-vbt {
-			filename = CONFIG_VBT_FILE;
-			offset = <CONFIG_VBT_ADDR>;
-		};
+	intel-vbt {
+		filename = CONFIG_VBT_FILE;
+		offset = <CONFIG_VBT_ADDR>;
+	};
 #endif
 #ifdef CONFIG_HAVE_REFCODE
-		intel-refcode {
-			offset = <CONFIG_X86_REFCODE_ADDR>;
-		};
+	intel-refcode {
+		offset = <CONFIG_X86_REFCODE_ADDR>;
+	};
 #endif
 #ifdef CONFIG_TPL
-		x86-start16-tpl {
-			offset = <CONFIG_SYS_X86_START16>;
-		};
+	x86-start16-tpl {
+		offset = <CONFIG_SYS_X86_START16>;
+	};
 #elif defined(CONFIG_SPL)
-		x86-start16-spl {
-			offset = <CONFIG_SYS_X86_START16>;
-		};
+	x86-start16-spl {
+		offset = <CONFIG_SYS_X86_START16>;
+	};
 #else
-		x86-start16 {
-			offset = <CONFIG_SYS_X86_START16>;
-		};
+	x86-start16 {
+		offset = <CONFIG_SYS_X86_START16>;
+	};
 #endif
 };
 #endif
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#93 lore raw

PATCH v2 44/50 x86: samus: Increase the pre-reloc memory again

In reply to Simon Glass

This is again too small, so increase it slightly.

---

Changes in v2: None

 configs/chromebook_samus_defconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
configs/chromebook_samus_defconfig +1 -1
index 7e2453f2440..5f5db07b5a5 100644
@@ -1,6 +1,6 @@
 CONFIG_X86=y
 CONFIG_SYS_TEXT_BASE=0xFFE00000
-CONFIG_SYS_MALLOC_F_LEN=0x1c00
+CONFIG_SYS_MALLOC_F_LEN=0x1d00
 CONFIG_DEBUG_UART_BOARD_INIT=y
 CONFIG_DEBUG_UART_BASE=0x3f8
 CONFIG_DEBUG_UART_CLOCK=1843200
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#94 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

In reply to Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

---

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
include/pci.h +1 -5
index 066238a9c3c..508f7bca81c 100644
@@ -546,11 +546,7 @@ extern void pci_cfgfunc_do_nothing(struct pci_controller* hose, pci_dev_t dev,
 extern void pci_cfgfunc_config_device(struct pci_controller* hose, pci_dev_t dev,
 				      struct pci_config_table *);
 
-#ifdef CONFIG_NR_DRAM_BANKS
-#define MAX_PCI_REGIONS (CONFIG_NR_DRAM_BANKS + 7)
-#else
-#define MAX_PCI_REGIONS 7
-#endif
+#define MAX_PCI_REGIONS		7
 
 #define INDIRECT_TYPE_NO_PCIE_LINK	1
 
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#95 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

+Thierry

Quoted 18 lines · Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

Quoted 8 lines

---

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
#96 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

Quoted 24 lines · Bin Meng

+Thierry

Quoted 21 lines · Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

---

Quoted 7 lines
Quoted 5 lines

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

Ugh... so we're trading one regression for another? Can we not live with the 2 KiB increase on x86 until this has been properly fixed? Currently this will cause Jetson TX2 to crash if it starts using PCI.

Thierry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20190502/41b3d8d6/attachment.sig>

#97 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

Hi Thierry,

Quoted 32 lines · Thierry Reding
Quoted 30 lines · Bin Meng

+Thierry

Quoted 26 lines · Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

---

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
Quoted 5 lines
Quoted 1 line

Ugh... so we're trading one regression for another? Can we not live with the 2 KiB increase on x86 until this has been properly fixed? Currently this will cause Jetson TX2 to crash if it starts using PCI.

Unfortunately this breaks several boards since we are out of memory.

I think this needs a better solution to reduce the memory usage down to sensible levels. This is something I should have considered when implementing the PCI uclass, but unfortunately I did not.

Regards,
Simon

#98 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

Hi Simon, Thierry,

Quoted 42 lines · Simon Glass

Hi Thierry,

Quoted 32 lines · Thierry Reding
Quoted 30 lines · Bin Meng

+Thierry

Quoted 26 lines · Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

---

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
Quoted 5 lines
Quoted 1 line

Ugh... so we're trading one regression for another? Can we not live with the 2 KiB increase on x86 until this has been properly fixed? Currently this will cause Jetson TX2 to crash if it starts using PCI.

Quoted 6 lines

Unfortunately this breaks several boards since we are out of memory.

I think this needs a better solution to reduce the memory usage down to sensible levels. This is something I should have considered when implementing the PCI uclass, but unfortunately I did not.

Could you please suggest whether I should apply this revert patch for now?

Regards,
Bin

#99 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

Hi Bin,

Quoted 46 lines · Bin Meng

Hi Simon, Thierry,

Quoted 42 lines · Simon Glass

Hi Thierry,

Quoted 37 lines · Thierry Reding
Quoted 24 lines · Bin Meng

+Thierry

Quoted 21 lines · Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

---

Quoted 7 lines
Quoted 5 lines

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

Ugh... so we're trading one regression for another? Can we not live with the 2 KiB increase on x86 until this has been properly fixed? Currently this will cause Jetson TX2 to crash if it starts using PCI.

Quoted 8 lines
Quoted 6 lines

Unfortunately this breaks several boards since we are out of memory.

I think this needs a better solution to reduce the memory usage down to sensible levels. This is something I should have considered when implementing the PCI uclass, but unfortunately I did not.

Could you please suggest whether I should apply this revert patch for now?

I suggest a temporary revert since this breaks some x86 boards.

I think the real fix is to reduce the memory used by PCI devices. Thierry, do you have time to look at this?

Regards,
Simon

#100 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

Hi Simon,

Quoted 57 lines · Simon Glass

Hi Bin,

Quoted 53 lines · Bin Meng

Hi Simon, Thierry,

Quoted 48 lines · Simon Glass

Hi Thierry,

Quoted 32 lines · Thierry Reding
Quoted 30 lines · Bin Meng

+Thierry

Quoted 26 lines · Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

---

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
Quoted 5 lines
Quoted 1 line

Ugh... so we're trading one regression for another? Can we not live with the 2 KiB increase on x86 until this has been properly fixed? Currently this will cause Jetson TX2 to crash if it starts using PCI.

Unfortunately this breaks several boards since we are out of memory.

I think this needs a better solution to reduce the memory usage down to sensible levels. This is something I should have considered when implementing the PCI uclass, but unfortunately I did not.

Quoted 6 lines
Quoted 1 line

Could you please suggest whether I should apply this revert patch for now?

I suggest a temporary revert since this breaks some x86 boards.

I think the real fix is to reduce the memory used by PCI devices. Thierry, do you have time to look at this?

OK, I will apply this patch then.

Regards,
Bin

#101 lore raw

PATCH v2 45/50 Revert "pci: Scale MAX_PCI_REGIONS based on CONFIG_NR_DRAM_BANKS"

In reply to Simon Glass
Quoted 56 lines · Simon Glass

Hi Bin,

Quoted 53 lines · Bin Meng

Hi Simon, Thierry,

Quoted 48 lines · Simon Glass

Hi Thierry,

Quoted 32 lines · Thierry Reding
Quoted 30 lines · Bin Meng

+Thierry

Quoted 26 lines · Simon Glass

This reverts commit aec4298ccb337106fd0115b91d846a022fdf301d.

Unfortunately this has a dramatic impact on the pre-relocation memory used on x86 platforms (increasing it by 2KB) since it increases the overhead for each PCI device from 220 bytes to 412 bytes.

The offending line is in UCLASS_DRIVER(pci):

.per_device_auto_alloc_size = sizeof(struct pci_controller),

This means that all PCI devices have the controller struct associated with them. The solution is to move the regions[] member out of the array, makes its size dynamic, or split UCLASS_PCI into controllers and non-controllers, as the comment suggests.

For now, revert the commit to get things running again.

---

Changes in v2: None

 include/pci.h | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
Quoted 5 lines
Quoted 1 line

Ugh... so we're trading one regression for another? Can we not live with the 2 KiB increase on x86 until this has been properly fixed? Currently this will cause Jetson TX2 to crash if it starts using PCI.

Unfortunately this breaks several boards since we are out of memory.

I think this needs a better solution to reduce the memory usage down to sensible levels. This is something I should have considered when implementing the PCI uclass, but unfortunately I did not.

Quoted 6 lines
Quoted 1 line

Could you please suggest whether I should apply this revert patch for now?

I suggest a temporary revert since this breaks some x86 boards.

I think the real fix is to reduce the memory used by PCI devices. Thierry, do you have time to look at this?

Not right away. I may get around to this within a couple of weeks maybe.

Thierry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20190509/16eaee49/attachment.sig>

#102 lore raw

PATCH v2 46/50 x86: Enable the RTC on all boards

In reply to Simon Glass

With the move to Kconfig this option should be set in Kconfig, not in the config header file. Move it.

---

Changes in v2:
- Add new patch to enable the RTC in Kconfig

 arch/Kconfig                      | 1 +
 configs/chromebook_link_defconfig | 1 +
 include/configs/x86-common.h      | 1 -
 3 files changed, 2 insertions(+), 1 deletion(-)
arch/Kconfig +1 -0
index a1d1ac301d6..2a93b72dbc4 100644
@@ -165,6 +165,7 @@ config X86
 	imply USB_ETHER_SMSC95XX
 	imply USB_HOST_ETHER
 	imply PCH
+	imply RTC_MC146818
 
 	# Thing to enable for when SPL/TPL are enabled: SPL
 	imply SPL_DM
configs/chromebook_link_defconfig +1 -0
index 6058dfaf0f2..b1c5095ec92 100644
@@ -53,6 +53,7 @@ CONFIG_DM_I2C=y
 CONFIG_SYS_I2C_INTEL=y
 CONFIG_CROS_EC=y
 CONFIG_CROS_EC_LPC=y
+CONFIG_RTC_MC146818=y
 CONFIG_SYS_NS16550=y
 CONFIG_SOUND=y
 CONFIG_SPI=y
include/configs/x86-common.h +0 -1
index 4180b25f977..7fcf76a6bf2 100644
@@ -36,7 +36,6 @@
 /*-----------------------------------------------------------------------
  * Real Time Clock Configuration
  */
-#define CONFIG_RTC_MC146818
 #define CONFIG_SYS_ISA_IO_BASE_ADDRESS	0
 #define CONFIG_SYS_ISA_IO      CONFIG_SYS_ISA_IO_BASE_ADDRESS
 
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#103 lore raw

PATCH v2 46/50 x86: Enable the RTC on all boards

Hi Simon,

Quoted 32 lines · Simon Glass

With the move to Kconfig this option should be set in Kconfig, not in the config header file. Move it.

---

Changes in v2:
- Add new patch to enable the RTC in Kconfig

 arch/Kconfig                      | 1 +
 configs/chromebook_link_defconfig | 1 +
 include/configs/x86-common.h      | 1 -
 3 files changed, 2 insertions(+), 1 deletion(-)
arch/Kconfig +1 -0
index a1d1ac301d6..2a93b72dbc4 100644
@@ -165,6 +165,7 @@ config X86
        imply USB_ETHER_SMSC95XX
        imply USB_HOST_ETHER
        imply PCH
+       imply RTC_MC146818
 
        # Thing to enable for when SPL/TPL are enabled: SPL
        imply SPL_DM
configs/chromebook_link_defconfig +0 -0
index 6058dfaf0f2..b1c5095ec92 100644

configs/chromebook_link_defconfig CONFIG_DM_I2C=y · line 56

Quoted 4 lines
configs/chromebook_link_defconfig +1 -0
@@ -53,6 +53,7 @@ CONFIG_DM_I2C=y
 CONFIG_SYS_I2C_INTEL=y
 CONFIG_CROS_EC=y
 CONFIG_CROS_EC_LPC=y
+CONFIG_RTC_MC146818=y

This has been implied in arch/Kconfig above, why adding another one here?

configs/chromebook_link_defconfig CONFIG_DM_I2C=y · line 59

Quoted 3 lines
configs/chromebook_link_defconfig +0 -0
 CONFIG_SYS_NS16550=y
 CONFIG_SOUND=y
 CONFIG_SPI=y

Regards,
Bin

#104 lore raw

PATCH v2 47/50 x86: Update the memory map a little

In reply to Simon Glass

The memory map currently omits the environment and the MRC region. Add these in for completeness.

---

Changes in v2: None

 doc/README.x86 | 2 ++
 1 file changed, 2 insertions(+)
doc/README.x86 +2 -0
index d5224b75367..8e0a3f36edf 100644
@@ -195,6 +195,8 @@ Flash map for samus / broadwell:
    ffed0000	SYS_TEXT_BASE
    ffea0000	X86_REFCODE_ADDR
    ffe70000	SPL_TEXT_BASE
+   ffbf8000	CONFIG_ENV_OFFSET (environemnt offset)
+   ffbe0000	rw-mrc-cache (Memory-reference-code cache)
    ffa00000	<spare>
    ff801000	intel-me (address set by descriptor.bin)
    ff800000	intel-descriptor
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#105 lore raw

PATCH v2 47/50 x86: Update the memory map a little

Hi Simon,

Quoted 4 lines · Simon Glass

The memory map currently omits the environment and the MRC region. Add these in for completeness.

I think the commit tile should be updated to add the "samus" tag.

Quoted 8 lines

---

Changes in v2: None

 doc/README.x86 | 2 ++
 1 file changed, 2 insertions(+)

Regards,
Bin

#106 lore raw

PATCH v2 48/50 x86: broadwell: Update PCH to work in TPL

In reply to Simon Glass

The early init should only happen once. Update the probe method to deal with TPL, SPL and U-Boot proper.

---

Changes in v2:
- Add a new patch to update PCH to work in TPL

 arch/x86/cpu/broadwell/pch.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
arch/x86/cpu/broadwell/pch.c +9 -3
index e61efa7b16c..a48945adf11 100644
@@ -599,10 +599,16 @@ static int broadwell_pch_init(struct udevice *dev)
 
 static int broadwell_pch_probe(struct udevice *dev)
 {
-	if (!(gd->flags & GD_FLG_RELOC))
-		return broadwell_pch_early_init(dev);
-	else
+	if (CONFIG_IS_ENABLED(X86_32BIT_INIT)) {
+		if (!(gd->flags & GD_FLG_RELOC))
+			return broadwell_pch_early_init(dev);
+		else
+			return broadwell_pch_init(dev);
+	} else if (IS_ENABLED(CONFIG_SPL) && !IS_ENABLED(CONFIG_SPL_BUILD)) {
 		return broadwell_pch_init(dev);
+	} else {
+		return 0;
+	}
 }
 
 static int broadwell_pch_get_spi_base(struct udevice *dev, ulong *sbasep)
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#107 lore raw

PATCH v2 48/50 x86: broadwell: Update PCH to work in TPL

Quoted 4 lines · Simon Glass

The early init should only happen once. Update the probe method to deal with TPL, SPL and U-Boot proper.

Quoted 9 lines

---

Changes in v2:
- Add a new patch to update PCH to work in TPL

 arch/x86/cpu/broadwell/pch.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
#108 lore raw

PATCH v2 49/50 x86: Add a way to jump from TPL to SPL

In reply to Simon Glass

When TPL finishes it needs to jump to SPL with the stack set up correctly. Add a function to handle this.

---

Changes in v2:
- Add a new patch allowing jumping from TPL to SPL

 arch/x86/cpu/start.S | 13 +++++++++++++
 1 file changed, 13 insertions(+)
arch/x86/cpu/start.S +13 -0
index 30fa7def464..4a82add76b7 100644
@@ -190,6 +190,19 @@ board_init_f_r_trampoline:
 	/* Re-enter U-Boot by calling board_init_f_r() */
 	call	board_init_f_r
 
+#ifdef CONFIG_TPL
+.globl jump_to_spl
+.type jump_to_spl, @function
+jump_to_spl:
+	/* Reset stack to the top of CAR space */
+	movl	$(CONFIG_SYS_CAR_ADDR + CONFIG_SYS_CAR_SIZE - 4), %esp
+#ifdef CONFIG_DCACHE_RAM_MRC_VAR_SIZE
+	subl	$CONFIG_DCACHE_RAM_MRC_VAR_SIZE, %esp
+#endif
+
+	jmp	*%eax
+#endif
+
 die:
 	hlt
 	jmp	die
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#109 lore raw

PATCH v2 49/50 x86: Add a way to jump from TPL to SPL

Hi Simon,

Quoted 4 lines · Simon Glass

When TPL finishes it needs to jump to SPL with the stack set up correctly. Add a function to handle this.

Quoted 9 lines

---

Changes in v2:
- Add a new patch allowing jumping from TPL to SPL

 arch/x86/cpu/start.S | 13 +++++++++++++
 1 file changed, 13 insertions(+)

But I don't see any patch in this series that makes use of this new jump_to_spl()?

Regards,
Bin

#110 lore raw

PATCH v2 49/50 x86: Add a way to jump from TPL to SPL

Hi Bin,

Quoted 14 lines · Bin Meng

Hi Simon,

Quoted 10 lines · Simon Glass

When TPL finishes it needs to jump to SPL with the stack set up correctly. Add a function to handle this.

---

Changes in v2:
- Add a new patch allowing jumping from TPL to SPL

Quoted 8 lines
Quoted 3 lines
 arch/x86/cpu/start.S | 13 +++++++++++++
 1 file changed, 13 insertions(+)

But I don't see any patch in this series that makes use of this new jump_to_spl()?

This is called from arch/x86/lib/tpl.c:

x86: Add a simple TPL implementation

Regards,
Simon

#111 lore raw

PATCH v2 50/50 x86: samus: Add a target to boot through TPL

In reply to Simon Glass

Add a version of samus which supports booting from TPL to SPL and then to U-Boot. This allows TPL to select from an A or B SPL to support verified boot with field upgrade.

---

Changes in v2:
- Sort defconfig and adjust it to build after rebase on maste

 board/google/Kconfig                      |  8 +++
 board/google/chromebook_samus/Kconfig     | 14 +++-
 board/google/chromebook_samus/MAINTAINERS |  7 ++
 configs/chromebook_samus_tpl_defconfig    | 81 +++++++++++++++++++++++
 include/configs/chromebook_samus.h        |  3 +
 5 files changed, 111 insertions(+), 2 deletions(-)
 create mode 100644 configs/chromebook_samus_tpl_defconfig
board/google/Kconfig +8 -0
index d98a5e818fc..679a0f10239 100644
@@ -52,6 +52,14 @@ config TARGET_CHROMEBOOK_SAMUS
 	  Chrome OS EC connected on LPC, and it provides a 2560x1700 high
 	  resolution touch-enabled LCD display.
 
+config TARGET_CHROMEBOOK_SAMUS_TPL
+	bool "Chromebook samus booting from TPL"
+	help
+	  This is a version of Samus which boots into TPL, then to SPL and
+	  U-Boot proper. This is useful where verified boot must select
+	  between different A/B versions of SPL/U-Boot, to allow upgrading of
+	  almost all U-Boot code in the field.
+
 endchoice
 
 source "board/google/chromebook_link/Kconfig"
board/google/chromebook_samus/Kconfig +12 -2
index afbfe53deb4..90c23cba1be 100644
@@ -1,4 +1,4 @@
-if TARGET_CHROMEBOOK_SAMUS
+if TARGET_CHROMEBOOK_SAMUS || TARGET_CHROMEBOOK_SAMUS_TPL
 
 config SYS_BOARD
 	default "chromebook_samus"
@@ -10,7 +10,8 @@ config SYS_SOC
 	default "broadwell"
 
 config SYS_CONFIG_NAME
-	default "chromebook_samus"
+	default "chromebook_samus" if TARGET_CHROMEBOOK_SAMUS
+	default "chromebook_samus" if TARGET_CHROMEBOOK_SAMUS_TPL
 
 config SYS_TEXT_BASE
 	default 0xffe00000
@@ -39,3 +40,12 @@ config SYS_CAR_SIZE
 	default 0x40000
 
 endif
+
+if TARGET_CHROMEBOOK_SAMUS_TPL
+
+config BOARD_SPECIFIC_OPTIONS_TPL # dummy
+	def_bool y
+	select SPL
+	select TPL
+
+endif
board/google/chromebook_samus/MAINTAINERS +7 -0
index 5500e46b408..ca4b16500af 100644
@@ -4,3 +4,10 @@ S:	Maintained
 F:	board/google/chromebook_samus/
 F:	include/configs/chromebook_samus.h
 F:	configs/chromebook_samus_defconfig
+
+CHROMEBOOK SAMUS TPL BOARD
+M:	Simon Glass <[email protected]>
+S:	Maintained
+F:	board/google/chromebook_samus/
+F:	include/configs/chromebook_samus.h
+F:	configs/chromebook_samus_tpl_defconfig
configs/chromebook_samus_tpl_defconfig new file +81 -0

81 lines not on this page · raw message

include/configs/chromebook_samus.h +3 -0
index ccb2fe8caad..610be1e7b5e 100644
@@ -23,4 +23,7 @@
 #define CONFIG_ENV_SECT_SIZE		0x1000
 #define CONFIG_ENV_OFFSET		0x003f8000
 
+#define CONFIG_SPL_TEXT_BASE		0xffe70000
+#define CONFIG_TPL_TEXT_BASE		0xfffd8000
+
 #endif	/* __CONFIG_H */
Signature · 1 line
2.21.0.593.g511ec345e18-goog
#112 lore raw

v2 cover x86: Add support for booting from TPL

In reply to Simon Glass

Hi Simon,

Quoted 12 lines · Simon Glass

At present SPL is used on 64-bit platforms, to allow SPL to be built as a 32-bit program and U-Boot proper to be built as 64-bit.

However it is useful to be able to use SPL on any x86 platform, where U-Boot needs to be updated in the field. Then SPL can select which U-Boot to run (A or B) and most of the code can be updated. Similarly, using TPL allows both SPL and U-Boot to be updated. This is the best approach, since it means that all of U-Boot proper as well as SPL (in particular SDRAM init) can be updated in the field. This provides for the smallest possible amount of read-only (non-updateable) code: just the TPL code.

Quoted 4 lines

This series contains a number of changes to allow x86 boards to use TPL, SPL and U-Boot proper. As a test, it is enabled for samus with a new chromebook_samus_tpl board.

I've applied 32 patches in this series to u-boot-x86/master. Remaining patches need to rebase on that.

Regards,
Bin