2
0
Fork 0
mirror of https://github.com/AsahiLinux/u-boot synced 2024-12-25 12:33:41 +00:00
Commit graph

2077 commits

Author SHA1 Message Date
Simon Glass
8ebca32b2d x86: Set the CPU vendor in SPL
We don't read this information in 64-bit mode, since we don't have the
macros for doing it. Set it to Intel by default. This allows the TSC timer
to work correctly.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
2c6b979ec1 x86: coreboot: Rearrange arch_cpu_init()
Init errors in SPL are currently ignored by this function.

Change the code to init the CPU, reporting an error if something is wrong.
After that, look for the coreboot table.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
7739b2793b x86: coreboot: Document cbmem console struct
Coreboot changed a few years ago to include an overflow flag. Update the
structure to match this.

This comes from coreboot commit:

   6f5ead14b4 ("mb/google/nissa/var/joxer: Update eMMC DLL settings")

Note: There are several implementations of this in coreboot. I have chosen
to follow the one in src/lib/cbmem_console.c

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
6acc072365 x86: dm: Mark driver model as dead when disabling CAR
When turning off CAR, set the flag to make sure that nothing tries to use
driver model in SPL before jumping to U-Bot proper, since its tables are
in CAR.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
cd1ee5d96e x86: broadwell: Set up MTRRs
The current condition does not handle the samus_tpl case where it sets
up the RAM in SPL but needs to commit the MTRRs in U-Boot proper.

Add another case to handle this and update the comment.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
3ca9b86e23 x86: broadwell: Avoid initing the CPU twice
When TPL has already set up the CPU, don't do it again. This existing
code actually has this backwards, so fix it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
623b3e8f97 x86: spl: Change the condition for copying U-Boot to RAM
Make this depend on whether the address matches the offset, rather than
a particular board build. For samus_tpl we don't need to copy, for
example.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
02de9199bc x86: Add some log categories
Add some missing log categories to a few files.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
8e34ddc113 x86: broadwell: Show the memory delay
Samus only takes 7 seconds but it is long enough to think it has hung. Add
a message about what it is doing, similar to the approach on coral.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Andy Shevchenko
b95bc64b06 x86: Prevent from missing the FADT chaining
Recent approach with FADT writer shows that there is
a room for subtle errors. Prevent this from happening
again by introducing acpi_add_fadt() helper.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
2e977b2ce8 Reland "x86: Move FACP table into separate functions""
Each board has its own way of creating this table. Rather than calling the
acpi_create_fadt() function for each one from a common acpi_write_fadt()
function, just move the writer into the board-specific code.

Co-developed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2023-09-22 06:03:46 +08:00
Troy Kisky
5c39f2c150 x86: cpu: i386: cpu: only set pci_ram_top if CONFIG_IS_ENABLED(PCI)
This avoids an error when ifdef CONFIG_PCI is changed to
if CONFIG_IS_ENABLED(PCI)

Signed-off-by: Troy Kisky <troykiskyboundary@gmail.com>
[Rebased on top of u-boot/master]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
2023-09-22 06:03:46 +08:00
Simon Glass
91caa3bb89 event: Use an event to replace last_stage_init()
Add a new event which handles this function. Convert existing use of
the function to use the new event instead.

Make sure that EVENT is enabled by affected boards, by selecting it from
the LAST_STAGE_INIT option. For x86, enable it by default since all boards
need it.

For controlcenterdc, inline the get_tpm() function and make sure the event
is not built in SPL.

Signed-off-by: Simon Glass <sjg@chromium.org>
2023-08-31 13:16:55 -04:00
Simon Glass
13a7db9ab1 x86: Convert arch_fsp_init() to use events
Convert this to use events instead of calling a function directly in the
init sequence.

Rename it to arch_fsp_init_f() to distinguish it from the one that happens
after relocation.

For FSPv2 nothing needs to be done here, so drop the empty function.

Signed-off-by: Simon Glass <sjg@chromium.org>
2023-08-31 13:16:54 -04:00
Simon Glass
f72d0d4a2f event: Convert existing spy records to simple
Very few of the existing event-spy records use the arguments they are
passed. Update them to use a simple spy instead, to simplify the code.

Where an adaptor function is currently used, remove it where possible.

Signed-off-by: Simon Glass <sjg@chromium.org>
2023-08-31 13:16:54 -04:00
Tom Rini
7e6e40c572 Prepare v2023.10-rc3
-----BEGIN PGP SIGNATURE-----
 
 iQGzBAABCgAdFiEEGjx/cOCPqxcHgJu/FHw5/5Y0tywFAmTjxwsACgkQFHw5/5Y0
 tyz4TAwAtVgllQGPVNIVDaEBdG93htm1DoZPc4EMZ4L5rf0nT0yJPeOhRCGygYY+
 zYL1ueleBZtpJhYOuHGetSns0B/M5Wx7gu1eGHNGrMNHH7zCV5Ynjjee2IaqKgbU
 tdSSc8XQpJ3PJY6rhJ6qZ7pO/EcUlQEdqplEUGFmA4JObx43j1+bmqyGTJA57vEw
 rrr+tyUVlr+Wv4lqlX/YJWzAB6IoSMUStv34M2lmiZvZoTYHezvIqBK33GrQ4Cwl
 hQ8nP7ymBXdA5GoMFBpfaBCA7aX3esx4yYN3JIRNRjNmnITJHYs0G8oIh8R5iNnJ
 psroU7GRPu3MmXfOnHlRfA/v0mo8xQGc2RQHyZUGXuApTI8+SFK/4WJPBny/IYse
 ggcVCHoI57dx1+j4+JHYhh61bMR1JUwARs4/UU4dKT3MRZM8Gb99OvbXO3+switZ
 R0YHSnDP///MTC7/MzbOPWzsJiTn9gkZd8nFVTXoncrVI8F3OtCc89G4HBruBSj7
 ptfylxqH
 =VlzD
 -----END PGP SIGNATURE-----

Merge tag 'v2023.10-rc3' into next

Prepare v2023.10-rc3

Signed-off-by: Tom Rini <trini@konsulko.com>
2023-08-21 17:32:17 -04:00
Heinrich Schuchardt
d768dd8855 common: return type board_get_usable_ram_top
board_get_usable_ram_top() returns a physical address that is stored in
gd->ram_top. The return type of the function should be phys_addr_t like the
current type of gd->ram_top.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
2023-08-15 18:21:17 +02:00
Simon Glass
cbb607d2d9 bootstd: Allow display of the x86 setup information
Provide an option to dump this information if available.

Move the funciion prototype to the common x86 header. Allow the command
line to be left out since 'bootflow info' show this itself and it is
not in the correct place in memory until the kernel is actually booted.

Fix a badly aligned heading while we are here.

Signed-off-by: Simon Glass <sjg@chromium.org>
2023-08-10 18:34:54 -06:00
Simon Glass
1a0810924a bootstd: Move common zimage functions to bootm.h
We want to avoid using #ifdefs around header files and in the code. It
makes sense to collect the various functions used for loading images into
a single header which can be included by all architectures. The best place
for this is the arch-neutral bootm.h header, so use that.

Move some zimage functions into this bootm.h header.

Signed-off-by: Simon Glass <sjg@chromium.org>
2023-08-10 18:34:54 -06:00
Simon Glass
3cc4095362 x86: qemu: Add required linux/sizes.h include
These files rely on the config.h file provided this include. Add it
explictily so we can move to a text environment.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> # Intel Edison
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-08-09 23:31:12 +08:00
Simon Glass
dbfb6c096e x86: i8254: Include required ibmpc.h header
This is needed for this file, so include it here explicitly.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> # Intel Edison
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-08-09 23:31:12 +08:00
Simon Glass
edd53bda53 x86: Drop CFG_SYS_STACK_SIZE
This is only used in one file and the value is the same for both boards
which define it. Use the fixed value of 32KB and drop the CFG. This will
allow removal of the config.h files.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> # Intel Edison
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-08-09 23:31:12 +08:00
Simon Glass
4099df48a6 x86: Correct copying of BIOS mode information
This is copying beyond the end of the destination buffer. Correct the code
by using the size of the  vesa_mode_info struct. We don't need to copy the
rest of the bytes in the buffer.

This long-standing bug prevents virtio bootdevs working correctly on
qemu-x86 at present.

Fixes: 0ca2426bea ("x86: Add support for running option ROMs natively")
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com> # qemu-x86_64
2023-08-09 23:31:11 +08:00
Simon Glass
1fa64e155d Revert "x86: Switch QEMU over to use the bochs driver"
Unfortunately the bochs driver does not currently work with distros.
It causes a hang between grub menu selection and the OS displaying
something.

Preliminary investigation shows that GRUB does not jump to the kernel
at all.

This reproduces reliably.

This reverts commit b8956425d5.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com> # qemu-x86_64
[Slightly modify the commit message about preliminary investigation]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
2023-08-09 23:31:11 +08:00
Simon Glass
ea6eef27ca x86: Run QEMU machine setup in SPL
Call the hardware-init function from QEMU from SPL. This allows the
video BIOS to operate correctly.

Create an x86-wide qemu.h header to avoid having to #ifdef the header
in spl.c

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com> # qemu-x86_64
2023-08-09 23:31:11 +08:00
Simon Glass
22080e05fc x86: spl: Drop unwanted debug()
This was left over from some previous debugging. Drop it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-08-09 23:31:11 +08:00
Shiji Yang
ccea96f443 treewide: unify the linker symbol reference format
Now all linker symbols are declared as type char[]. Though we can
reference the address via both the array name 'var' and its address
'&var'. It's better to unify them to avoid confusing developers.
This patch converts all '&var' linker symbol refrences to the most
commonly used format 'var'.

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
2023-08-09 09:21:42 -04:00
Shiji Yang
506df9dc58 treewide: rework linker symbol declarations in sections header
1. Convert all linker symbols to char[] type so that we can get the
   corresponding address by calling array name 'var' or its address
   '&var'. In this way, we can avoid some potential issues[1].
2. Remove unused symbol '_TEXT_BASE'. It has been abandoned and has
   not been referenced by any source code.
3. Move '__data_end' to the arch x86's own sections header as it's
   only used by x86 arch.
4. Remove some duplicate declared linker symbols. Now we use the
   standard header file to declare them.

[1] This patch fixes the boot failure on MIPS target. Error log:
SPL: Image overlaps SPL

Fixes: 1b8a1be1a1 ("spl: spl_legacy: Fix spl_end address")
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
2023-08-09 09:21:42 -04:00
Simon Glass
db971a7587 x86: Add a little more info to cbsysinfo
Show the number of records in the table and the total table size in
bytes.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-08-01 10:08:49 +08:00
Simon Glass
d560f7cae0 x86: Return mtrr_add_request() to its old purpose
This function used to be for adding a list of requests to be actioned on
relocation. Revert it back to this purpose, to avoid problems with boards
which need control of their MTRRs (i.e. those which don't use FSP).

The mtrr_set_next_var() function is available when the next free
variable-MTRR must be set, so this can be used instead.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Fixes: 3bcd6cf89e ("x86: mtrr: Skip MSRs that were already programmed..")
Fixes: 596bd0589a ("x86: mtrr: Do not clear the unused ones..")
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
2023-08-01 10:06:46 +08:00
Bin Meng
3dfa411501 x86: fsp: Use mtrr_set_next_var() for graphics memory
At present this uses mtrr_add_request() & mtrr_commit() combination
to program the MTRR for graphics memory. This usage has two major
issues as below:

- mtrr_commit() will re-initialize all MTRR registers from index 0,
  using the settings previously added by mtrr_add_request() and saved
  in gd->arch.mtrr_req[], which won't cause any issue but is unnecessary
- The way such combination works is based on the assumption that U-Boot
  has full control with MTRR programming (e.g.: U-Boot without any blob
  that does all low-level initialization on its own, or using FSP2 which
  does not touch MTRR), but this is not the case with FSP. FSP programs
  some MTRRs during its execution but U-Boot does not have the settings
  saved in gd->arch.mtrr_req[] and when doing mtrr_commit() it will
  corrupt what was already programmed previously.

Correct this to use mtrr_set_next_var() instead.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
2023-08-01 10:06:46 +08:00
Bin Meng
41fbb34469 x86: Change testing logic of mtrr commit
On Coral U-Boot SPL programs some MTRRs and FSPv2 in U-Boot proper
needs to program MTRRs too. With current testing logic of mtrr
commit in init_cache_f_r(), the mtrr commit is skipped which won't
work as the queued mtrr requests include setup for DRAM regions.

Change the logic to allow such configuration.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Tweak to put back CONFIG_FSP_VERSION2 at top:
Signed-off-by: Simon Glass <sjg@chromium.org>
2023-08-01 10:06:46 +08:00
Simon Glass
b8956425d5 x86: Switch QEMU over to use the bochs driver
This is more convenient since it does not require a video BIOS. Enable
it for QEMU.

Also drop use of video in SPL for the 64-bit QEMU, since it not needed
now.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:23:15 +08:00
Simon Glass
dac1fa5c19 x86: Make sure that the LPC is active before SDRAM init
Some boards need to access GPIOs to determine which SDRAM is fitted to the
board, for example chromebook_link. Probe this device (if it exists) to
make sure that this works as expected.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:23:15 +08:00
Simon Glass
df1bb2cb0b x86: link: Support Micron memory
Add the required tag so that micron memory can be set up correctly.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:23:15 +08:00
Simon Glass
e2e7de8747 x86: Convert some debug statements to use logging
Move from using debug() to log_debug() so that we don't have to use the
__func__ parameter and can access other logging features.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:23:14 +08:00
Simon Glass
6a32489782 x86: Record the start and end of the tables
The ACPI tables are special in that they are passed to EFI as a separate
piece, independent of other tables.

Also they can be spread over two areas of memory, e.g. with QEMU we end
up with tables kept in high memory as well.

Add new global_data fields to hold this information and update the bdinfo
command to show the table areas.

Move the rom_table_end variable into the loop that uses it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:23:08 +08:00
Simon Glass
8856d613cb x86: Refactor table-writing code a little
The implementation of write_tables() is confusing because it uses the
rom_table_start variable as the address pointer as it progresses.

Rename it to rom_addr to make the code clearer. Move the rom_table_end
variable into the block where it is used.

Also update logging to use the ACPI category, now that it is available.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:12:26 +08:00
Simon Glass
e7595aa350 x86: Allow logging to be used in SPL reliably
When global_data is relocated, log_head moves in memory, meaning that
the items in that list point to the wrong place.

Disable logging when making the change, then reenable it afterwards, so
that logging works normally.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:12:26 +08:00
Simon Glass
7c10e111c1 x86: Init video in SPL if enabled
When video is required in SPL, set this up ready for use. Ignore any
problems since it may be that video is not actually available and we
still want to continue on to U-Boot proper in that case.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:12:26 +08:00
Simon Glass
3525258019 x86: Ensure SPL banner is only shown once
Print the banner in SPL init only if the spl_board_init() function isn't
enabled. The spl_board_init() function is in the same file, but is called
later, by board_init_r().

This avoids printing two banners, which causes tests to fail.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:12:26 +08:00
Simon Glass
5345700d2a x86: Allow video-BIOS code to be built for SPL
With qemu-x86_64 we need to run the video BIOS while in 32-bit mode, i.e.
SPL. Add a Kconfig option for this, adjust the Makefile rules and use
CONFIG_IS_ENABLED() where needed.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:12:26 +08:00
Simon Glass
8f015d3781 x86: Improve the trampoline in 64-bit mode
At present this leaves the stack at the pre-relocation value. This is not
ideal since we want to have U-Boot running entirely from the top of
memory.

In addition, the new global_data pointer is not actually used, since
the global_data pointer itself is relocated, then the pre-relocation value
is changed, so the effective value (after relocation) does not update.

Adjust the implementation to follow the 32-bit code more closely, with a
trampoline function which is passed the new stack and global_data pointer.
This ensures that the correct values come through even when relocating.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:12:07 +08:00
Simon Glass
78f24d8f34 x86: Show an error when a BIOS exception occurs
Rather than silently hanging, show an error first. This can happen when
there is something wrong with the video BIOS.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:08:44 +08:00
Simon Glass
67884002f4 x86: Correct get_sp() implementation for 64-bit
Use an assembler implementation as is done for i386, so that the results
are equivalent for i386 and x86_64.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:08:44 +08:00
Simon Glass
633af11dd6 x86: Show the CPU physical address size with bdinfo
This is useful information so show it with the bdinfo command.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:08:44 +08:00
Simon Glass
de94db8132 x86: Add a comment for board_init_f_r_trampoline()
Add a comment for this function in the header.

Change the function (and the one after) to use __noreturn to keep
checkpatch happy.

Add docs to board_init_f_r() while we are here.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:08:44 +08:00
Simon Glass
b73dba7a83 x86: Tidy up EFI code in interrupt_init()
The ll_boot_init() check handles the EFI case so we don't need the rest
of the code. Drop it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:08:44 +08:00
Simon Glass
4fb2536e5b x86: Allow listing MTRRs in SPL
Move MTRR-listing code into a common file so it can be used from SPL.
Update the 'mtrr' command to call it.

Use this in SPL just before adjusting the MTRRs, so we can see the state
set up by the board. Only show it when debug is enabled.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2023-07-17 17:08:44 +08:00
Simon Glass
d0dfbf548d x86: zimage: Export the function to obtain the cmdline
Allow reading the command line from a zimage, so that it can be recorded
in the bootflow.

Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
2023-07-17 13:38:35 +08:00