NuKVM/memtest86plus
3
0
Fork 0
mirror of https://github.com/memtest86plus/memtest86plus.git synced 2024-11-27 10:00:17 -06:00
Code Issues Packages Projects Releases Wiki Activity
240 Commits 34 Branches 9 Tags 3.1 MiB
2a7631daab
Commit Graph

240 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Sam Demeulemeester
5249ae0af7
Merge pull request #27 from memtest86plus/usb_size_optimization 
Small optimizations in the EHCI code
 2022-03-31 01:10:35 +02:00
Lionel Debroux
e0d42db07c Optimize the code by declaring several functions static in system/ehci.c. 
`size memtest_shared` indicates that this saves 200- (x86) / 200+ (x86_64) bytes, but alignment hides the gain.
 2022-03-30 11:54:56 +02:00
Sam Demeulemeester
b6e2a2ace8 Attempt to fix another issue related to #19 by adding a BENCH_MIN_START_ADR constant to force bench location > 16MB in all cases. Correct mem_test_len by adding a x2 multiplier (len=src+dst) 2022-03-29 00:42:34 +02:00
Sam Demeulemeester
a4f1ba37b3 Add support for AMD Ryzen FCH (#21) 2022-03-27 23:47:57 +02:00
Sam Demeulemeester
f7ae4dd395 Add support for Zhaoxin ZX-E & KX CPUs 2022-03-25 20:14:16 +01:00
Sam Demeulemeester
88017f007f Add preliminary support for XMP3 on DDR5 (#23). Fix an issue with wrong SPD values on DDR5 ES modules 2022-03-25 00:33:42 +01:00
Sam Demeulemeester
aaa2061aec Fix an issue with DDR5 capacity detection 2022-03-24 21:49:56 +01:00
Sam Demeulemeester
b22c032b5a Add support for DDR5 in smbus.c. Major rework of various related functions (#2) 2022-03-24 21:49:56 +01:00
Sam Demeulemeester
8e35753ce8 Fix attempt for benchmark memory allocation (#19) 2022-03-24 21:49:56 +01:00
Sam Demeulemeester
b4bab4f829 Add decoding of DMI structure 17 as helper for SPD decoding. Move DMI display function to disaply.h 2022-03-24 21:49:56 +01:00
Sam Demeulemeester
d38df135db Update README.md before going public 2022-03-24 21:49:56 +01:00
Sam Demeulemeester
fa206bb865 Added DDR2 SPD decoding (#2) 2022-03-24 21:49:56 +01:00
Sam Demeulemeester
ec1a880288 Reset methods re-ordering (proven more universal for BIOS/UEFI as an attempt to solve #17) 2022-03-24 21:49:56 +01:00
Sam Demeulemeester
bd5e6f3c6b Bring back the cache & memory benchmark feature (preliminary) 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
6e57ffec34 Added a lot of modern IMC from Intel's BDW (5th gen) to RPL (13th gen) 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
eaf1cbeebb Minor cosmectic changes on smbus.c & Added Amazon JEP106 per AWS team request 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
04c3c485a1 Minor fixes on previously added temperature code 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
f9690813ca New code base for SMBUS / SPD access (#2) - WIP 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
de55c0428c Readd hwctrl.h & add argument for reset type 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
2266151fe6 Early SMBUS functions (#2) & EFI Reset (#17) 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
b12186b0aa Code cleanup & reindent on smbios.c (#1) 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
63904403b2 Added preliminary BIOS/EFI SMBIOS table parsing & basic printing (#1) 2022-03-24 21:49:56 +01:00
Sam DEMEULEMEESTER
300fd7c1cf Add temperature display for AMD Zen CPU (#3) 2022-03-24 21:49:56 +01:00
Martin Whitaker
8e600cd9f6 Include our version of string.h, not the build system one. 2022-03-11 18:07:39 +00:00
Martin Whitaker
0873e3b60a Include string.h, not memory.h, when using memcpy and memset. 
memory.h is an anachronism, and not present on all systems.
 2022-03-11 18:07:01 +00:00
Martin Whitaker
f6da06b117 Work around the halt/wakeup race in barrier_halt_wait() (issue #13) 
There is an unavoidable race between one core halting after decrementing
the barrier count and another core sending it the wakeup NMI. This can
only occur if the core sending the wakeup is running at many times the
speed of the core halting, but it has been observed on an Intel Icelake
mobile processor.
 2022-03-08 23:25:16 +00:00
Martin Whitaker
0076e63885 Optimise the barrier wait code. 
This reduces the number of instructions between decrementing the count
and halting in the halt wait case. Use the same code for the spin wait
case for consistency.
 2022-03-08 23:25:16 +00:00
Martin Whitaker
ed0fd7830f Use IP instead of PC in interrupt display. 
It's better to use the x86 terminology here.
 2022-03-08 23:25:16 +00:00
Doc TB
3ead0458c5
Merge pull request #16 from memtest86plus/x86fr/misc_features 
New reset method
 2022-03-08 23:42:23 +01:00
Sam DEMEULEMEESTER
4bc3b6a3c8 Add additional reset method 2022-03-08 23:30:31 +01:00
Sam DEMEULEMEESTER
09f5525f33 Add additional reset method 2022-03-08 23:24:25 +01:00
Martin Whitaker
b609c4dd81 Make detection of EHCI asynchronous transfer complete more robust. 
Using Interrupt On Completion is not robust, because the interrupt flag
is also set if a short transfer is detected. So we need to poll the
Active flag in the transfer descriptors.
 2022-03-06 18:20:53 +00:00
Martin Whitaker
e92f488753 Improve efficiency of random number generation (discussion #8). 
Use a more efficient algorithm that can be in-lined, and keep the
generator state in a local variable.
 2022-03-05 20:04:32 +00:00
Martin Whitaker
5e2ab9289b Make sure the XHCI private memory is accessible. 
The XHCI device index and scratchpad buffers are mapped into high memory,
in order to conserve low memory. They need to be accessible in the virtual
address space to allow us to initialise them. After initialisation. only
the XHCI accesses them.

We also only need to access the data structures in the device private
workspace during initialisation, but keeping separate physical and virtual
addresses for these structures makes the code considerably more complex,
so for now, move these to low memory.
 2022-03-05 16:27:24 +00:00
Martin Whitaker
6c79f4d214 Fix communication with low speed devices in OHCI driver (issue #6). 2022-03-04 18:15:17 +00:00
Martin Whitaker
4078b7760e Faster barrier implementation. 
The old barrier implementation was very slow when running on a multi-socket
machine (pcmemtest issue 16).

The new implementation provides two options:

  - when blocked, spin on a thread-local flag
  - when blocked, execute a HLT instruction and wait for a NMI

The first option might be faster, but we need to measure it to find out. A
new boot command line option is provided to select between the two, with a
third setting that uses a mixture of the two.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
311a597766 Remove unnecessary memory fence from spin_lock(). 
The x86 locked instructions enforce total memory ordering, so the
additional fence is redundant.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
b3d350f454 Add support for 64-bit local APIC addresses found in the MADT. 
This is untested, as I don't have any hardware that uses this option.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
631cd553aa Make sure the cache is enabled on all CPU cores, not just the BSP. 2022-02-28 22:05:21 +00:00
Martin Whitaker
06d670e9a8 Make sure the boot command line is accessible during startup. 
If we remap other regions, it's possible it could get overlaid.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
03464dce85 Make sure the bootparams structure remains accessible during startup. 
Once we start remapping other regions, it's possible it could get overlaid.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
9fca23105a Don't search for the RDSP at bootparams.acpi_rsdp_addr. 
This field is only used internally by the Linux kernel, so will always
be zero when bootparams is passed to us by a bootloader.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
da7829d4c3 Ensure the ACPI tables and local APIC are mapped into an accessible region. 
We only map the first 4GB of physical address space, so if the ACPI tables
or local APIC are located above 4GB, or are overlaid when we remap something
else (e.g. the video frame buffer), we need to map them to somewhere we can
access. The ACPI tables are only used during startup, but the local APIC
will be needed when we are running tests if we are saving power by halting
idle CPU cores and using a NMI to wake them up.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
5211b67e6c Add option to map_region to indicate we only use the region during startup. 
This will avoid unnecessary remapping of the ACPI tables if they are
located in the third GB of memory.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
a65d6a795d Rename map_device to map_region to better describe its use. 2022-02-28 22:05:21 +00:00
Martin Whitaker
604b22449a Allow device mapping to reuse existing mapped regions when they fully overlap. 2022-02-28 22:05:21 +00:00
Martin Whitaker
e4ad2dfa1b Don't use the top 0.5GB of the 32-bit address space for device mapping. 
This region is commonly used by the BIOS for mapping hardware devices,
so we can avoid a lot of remapping if we don't use it.
 2022-02-28 22:05:21 +00:00
Martin Whitaker
6b570cd779 Reduce the maximum mapped size of the linear frame buffer. 
This is needed for subsequent changes. If we do ever get presented with
a frame buffer larger than 8192x8192 pixels, we'll need to think again
about how to manage it.
 2022-02-28 22:05:21 +00:00
martinwhitaker
15b233cc89
Merge pull request #4 from memtest86plus/debrouxl_misc_improvements 
Misc small improvements
 2022-02-22 23:17:05 +00:00
Lionel Debroux
3ed9276eb0 Optimize the interrupt handler a bit by using compile-time string concatenation. 2022-02-22 11:00:02 +01:00