cachepc-linux

Fork of AMDESE/linux with modifications for CachePC side-channel attack
git clone https://git.sinitax.com/sinitax/cachepc-linux
Log | Files | Refs | README | LICENSE | sfeed.txt

ramoops.rst (6384B)


      1Ramoops oops/panic logger
      2=========================
      3
      4Sergiu Iordache <sergiu@chromium.org>
      5
      6Updated: 10 Feb 2021
      7
      8Introduction
      9------------
     10
     11Ramoops is an oops/panic logger that writes its logs to RAM before the system
     12crashes. It works by logging oopses and panics in a circular buffer. Ramoops
     13needs a system with persistent RAM so that the content of that area can
     14survive after a restart.
     15
     16Ramoops concepts
     17----------------
     18
     19Ramoops uses a predefined memory area to store the dump. The start and size
     20and type of the memory area are set using three variables:
     21
     22  * ``mem_address`` for the start
     23  * ``mem_size`` for the size. The memory size will be rounded down to a
     24    power of two.
     25  * ``mem_type`` to specify if the memory type (default is pgprot_writecombine).
     26
     27Typically the default value of ``mem_type=0`` should be used as that sets the pstore
     28mapping to pgprot_writecombine. Setting ``mem_type=1`` attempts to use
     29``pgprot_noncached``, which only works on some platforms. This is because pstore
     30depends on atomic operations. At least on ARM, pgprot_noncached causes the
     31memory to be mapped strongly ordered, and atomic operations on strongly ordered
     32memory are implementation defined, and won't work on many ARMs such as omaps.
     33Setting ``mem_type=2`` attempts to treat the memory region as normal memory,
     34which enables full cache on it. This can improve the performance.
     35
     36The memory area is divided into ``record_size`` chunks (also rounded down to
     37power of two) and each kmesg dump writes a ``record_size`` chunk of
     38information.
     39
     40Limiting which kinds of kmsg dumps are stored can be controlled via
     41the ``max_reason`` value, as defined in include/linux/kmsg_dump.h's
     42``enum kmsg_dump_reason``. For example, to store both Oopses and Panics,
     43``max_reason`` should be set to 2 (KMSG_DUMP_OOPS), to store only Panics
     44``max_reason`` should be set to 1 (KMSG_DUMP_PANIC). Setting this to 0
     45(KMSG_DUMP_UNDEF), means the reason filtering will be controlled by the
     46``printk.always_kmsg_dump`` boot param: if unset, it'll be KMSG_DUMP_OOPS,
     47otherwise KMSG_DUMP_MAX.
     48
     49The module uses a counter to record multiple dumps but the counter gets reset
     50on restart (i.e. new dumps after the restart will overwrite old ones).
     51
     52Ramoops also supports software ECC protection of persistent memory regions.
     53This might be useful when a hardware reset was used to bring the machine back
     54to life (i.e. a watchdog triggered). In such cases, RAM may be somewhat
     55corrupt, but usually it is restorable.
     56
     57Setting the parameters
     58----------------------
     59
     60Setting the ramoops parameters can be done in several different manners:
     61
     62 A. Use the module parameters (which have the names of the variables described
     63 as before). For quick debugging, you can also reserve parts of memory during
     64 boot and then use the reserved memory for ramoops. For example, assuming a
     65 machine with > 128 MB of memory, the following kernel command line will tell
     66 the kernel to use only the first 128 MB of memory, and place ECC-protected
     67 ramoops region at 128 MB boundary::
     68
     69	mem=128M ramoops.mem_address=0x8000000 ramoops.ecc=1
     70
     71 B. Use Device Tree bindings, as described in
     72 ``Documentation/devicetree/bindings/reserved-memory/ramoops.yaml``.
     73 For example::
     74
     75	reserved-memory {
     76		#address-cells = <2>;
     77		#size-cells = <2>;
     78		ranges;
     79
     80		ramoops@8f000000 {
     81			compatible = "ramoops";
     82			reg = <0 0x8f000000 0 0x100000>;
     83			record-size = <0x4000>;
     84			console-size = <0x4000>;
     85		};
     86	};
     87
     88 C. Use a platform device and set the platform data. The parameters can then
     89 be set through that platform data. An example of doing that is:
     90
     91 .. code-block:: c
     92
     93  #include <linux/pstore_ram.h>
     94  [...]
     95
     96  static struct ramoops_platform_data ramoops_data = {
     97        .mem_size               = <...>,
     98        .mem_address            = <...>,
     99        .mem_type               = <...>,
    100        .record_size            = <...>,
    101        .max_reason             = <...>,
    102        .ecc                    = <...>,
    103  };
    104
    105  static struct platform_device ramoops_dev = {
    106        .name = "ramoops",
    107        .dev = {
    108                .platform_data = &ramoops_data,
    109        },
    110  };
    111
    112  [... inside a function ...]
    113  int ret;
    114
    115  ret = platform_device_register(&ramoops_dev);
    116  if (ret) {
    117	printk(KERN_ERR "unable to register platform device\n");
    118	return ret;
    119  }
    120
    121You can specify either RAM memory or peripheral devices' memory. However, when
    122specifying RAM, be sure to reserve the memory by issuing memblock_reserve()
    123very early in the architecture code, e.g.::
    124
    125	#include <linux/memblock.h>
    126
    127	memblock_reserve(ramoops_data.mem_address, ramoops_data.mem_size);
    128
    129Dump format
    130-----------
    131
    132The data dump begins with a header, currently defined as ``====`` followed by a
    133timestamp and a new line. The dump then continues with the actual data.
    134
    135Reading the data
    136----------------
    137
    138The dump data can be read from the pstore filesystem. The format for these
    139files is ``dmesg-ramoops-N``, where N is the record number in memory. To delete
    140a stored record from RAM, simply unlink the respective pstore file.
    141
    142Persistent function tracing
    143---------------------------
    144
    145Persistent function tracing might be useful for debugging software or hardware
    146related hangs. The functions call chain log is stored in a ``ftrace-ramoops``
    147file. Here is an example of usage::
    148
    149 # mount -t debugfs debugfs /sys/kernel/debug/
    150 # echo 1 > /sys/kernel/debug/pstore/record_ftrace
    151 # reboot -f
    152 [...]
    153 # mount -t pstore pstore /mnt/
    154 # tail /mnt/ftrace-ramoops
    155 0 ffffffff8101ea64  ffffffff8101bcda  native_apic_mem_read <- disconnect_bsp_APIC+0x6a/0xc0
    156 0 ffffffff8101ea44  ffffffff8101bcf6  native_apic_mem_write <- disconnect_bsp_APIC+0x86/0xc0
    157 0 ffffffff81020084  ffffffff8101a4b5  hpet_disable <- native_machine_shutdown+0x75/0x90
    158 0 ffffffff81005f94  ffffffff8101a4bb  iommu_shutdown_noop <- native_machine_shutdown+0x7b/0x90
    159 0 ffffffff8101a6a1  ffffffff8101a437  native_machine_emergency_restart <- native_machine_restart+0x37/0x40
    160 0 ffffffff811f9876  ffffffff8101a73a  acpi_reboot <- native_machine_emergency_restart+0xaa/0x1e0
    161 0 ffffffff8101a514  ffffffff8101a772  mach_reboot_fixups <- native_machine_emergency_restart+0xe2/0x1e0
    162 0 ffffffff811d9c54  ffffffff8101a7a0  __const_udelay <- native_machine_emergency_restart+0x110/0x1e0
    163 0 ffffffff811d9c34  ffffffff811d9c80  __delay <- __const_udelay+0x30/0x40
    164 0 ffffffff811d9d14  ffffffff811d9c3f  delay_tsc <- __delay+0xf/0x20