When a driver declares DM_FLAG_PRE_RELOC flag, it wishes to be
bound before relocation. However due to a bug in the DM core,
the flag only takes effect when devices are statically declared
via U_BOOT_DEVICE(). This bug has been fixed recently by commit
"dm: core: Respect drivers with the DM_FLAG_PRE_RELOC flag in
lists_bind_fdt()", but with the fix, it has a side effect that
all existing drivers that declared DM_FLAG_PRE_RELOC flag will
be bound before relocation now. This may expose potential boot
failure on some boards due to insufficient memory during the
pre-relocation stage.
To mitigate this potential impact, the following changes are
implemented:
- Remove DM_FLAG_PRE_RELOC flag in the driver, if the driver
only supports configuration from device tree (OF_CONTROL)
- Keep DM_FLAG_PRE_RELOC flag in the driver only if the device
is statically declared via U_BOOT_DEVICE()
- Surround DM_FLAG_PRE_RELOC flag with OF_CONTROL check, for
drivers that support both statically declared devices and
configuration from device tree
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
So far the TSC timer driver supports trying hardware calibration first
and using device tree as last resort for its running frequency as the
normal timer.
However when it is used as the early timer, it only supports hardware
calibration and if it fails, the driver just panics. This introduces
a new config option to specify the early timer frequency in MHz and
it should be equal to the value described in the device tree.
Without this patch, the travis-ci testing on QEMU x86_64 target fails
each time after it finishes the 'bootefi selftest' as the test.py see
an error was emitted on the console like this:
TSC frequency is ZERO
resetting ...
### ERROR ### Please RESET the board ###
It's strange that this error is consistently seen on the travis-ci
machine, but only occasionally seen on my local machine (maybe 1 out
of 10). Since QEMU x86_64 target enables BOOTSTAGE support which uses
early timer, with this fix it should work without any failure.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
At present if TSC frequency is provided in the device tree, it takes
precedence over hardware calibration result. This swaps the order to
try hardware calibration first and uses device tree as last resort.
This can be helpful when a generic dts (eg: coreboot/efi payload) is
supposed to work on as many hardware as possible, including emulators
like QEMU where TSC hardware calibration sometimes fails.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
With the introduction of early timer support in the TSC driver,
the capability of getting clock rate from device tree was lost
unfortunately. Now we bring such functionality back, but with a
limitation that when TSC is used as early timer, specifying clock
rate from device tree does not work.
This fixes random boot failures seen on QEMU targets: printing "TSC
frequency is ZERO" and reset forever.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Starting with cpuid level 0x16 (Skylake-based processors)
it is possible to get CPU base freq via cpuid.
This fixes booting on a skylake based system.
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: fixed wrong indention of labels]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <trini@konsulko.com>
With bootstage we need access to the timer before driver model is set up.
To handle this, put the required state in global_data and provide a new
function to set up the device, separate from the driver's probe() method.
This will be used by the 'early' timer also.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Per the Intel 64 and IA-32 Architecture Software Developer's Manual,
add the reference clock for Intel Atom Processors based on the Airmont
Microarchitecture (Braswell).
This keeps in sync with Linux kernel commit:
6fcb41c: x86/tsc_msr: Add Airmont reference clock values
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Rename try_msr_calibrate_tsc() to cpu_mhz_from_msr(), as that
better describes what the routine does.
This keeps in sync with Linux kernel commit:
02c0cd2: x86/tsc_msr: Remove irqoff around MSR-based TSC enumeration
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Atom processors use a 19.2 MHz crystal oscillator.
Early processors generate 100 MHz via 19.2 MHz * 26 / 5 = 99.84 MHz.
Later processors generate 100 MHz via 19.2 MHz * 125 / 24 = 100 MHz.
Update the Silvermont-based tables accordingly, matching the Software
Developers Manual.
Also, correct a 166 MHz entry that should have been 116 MHz, and add
a missing 80 MHz entry for VLV2.
This keeps in sync with Linux kernel commit:
05680e7: x86/tsc_msr: Correct Silvermont reference clock values
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Some processor abbreviations in the comments of freq_desc_tables[]
are obscure. This updates part of these to mention processors
that are known to us. Also expand frequency definitions.
This keeps in sync with Linux kernel commit:
9e0cae9: x86/tsc_msr: Update comments, expand definitions
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
If either ratio or freq is zero, the return value is zero. There
is no need to create a fail branch and return zero there.
This keeps in sync with Linux kernel commit:
14bb4e3: x86/tsc_msr: Remove debugging messages
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
try_msr_calibrate_tsc() is currently Intel-specific, and should not
execute on any other vendor's parts.
This keeps in sync with Linux kernel commit:
ba82683: x86/tsc_msr: Identify Intel-specific code
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Currently we read the tsc radio like this:
ratio = (MSR_PLATFORM_INFO >> 8) & 0x1f;
Thus we get bit 8-12 of MSR_PLATFORM_INFO, however according to the
Intel manual, the ratio bits are bit 8-15.
Fix this problem by masking 0xff instead.
This keeps in sync with Linux kernel commit:
886123f: x86/tsc: Read all ratio bits from MSR_PLATFORM_INFO
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Right now i8254_init() is called from timer_init() in the tsc timer
driver. But actually i8254 and tsc are completely different things.
Since tsc timer has been converted to driver model, we should find
a new place that is appropriate for U-Boot to call i8254_init(),
which is now x86_cpu_init_f().
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
To group all dm timer drivers together, move tsc timer to
drivers/timer directory.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
