1Ramoops oops/panic logger 2========================= 3 4Sergiu Iordache <sergiu@chromium.org> 5 6Updated: 17 November 2011 7 80. Introduction 9 10Ramoops is an oops/panic logger that writes its logs to RAM before the system 11crashes. It works by logging oopses and panics in a circular buffer. Ramoops 12needs a system with persistent RAM so that the content of that area can 13survive after a restart. 14 151. Ramoops concepts 16 17Ramoops uses a predefined memory area to store the dump. The start and size 18and type of the memory area are set using three variables: 19 * "mem_address" for the start 20 * "mem_size" for the size. The memory size will be rounded down to a 21 power of two. 22 * "mem_type" to specifiy if the memory type (default is pgprot_writecombine). 23 24Typically the default value of mem_type=0 should be used as that sets the pstore 25mapping to pgprot_writecombine. Setting mem_type=1 attempts to use 26pgprot_noncached, which only works on some platforms. This is because pstore 27depends on atomic operations. At least on ARM, pgprot_noncached causes the 28memory to be mapped strongly ordered, and atomic operations on strongly ordered 29memory are implementation defined, and won't work on many ARMs such as omaps. 30 31The memory area is divided into "record_size" chunks (also rounded down to 32power of two) and each oops/panic writes a "record_size" chunk of 33information. 34 35Dumping both oopses and panics can be done by setting 1 in the "dump_oops" 36variable while setting 0 in that variable dumps only the panics. 37 38The module uses a counter to record multiple dumps but the counter gets reset 39on restart (i.e. new dumps after the restart will overwrite old ones). 40 41Ramoops also supports software ECC protection of persistent memory regions. 42This might be useful when a hardware reset was used to bring the machine back 43to life (i.e. a watchdog triggered). In such cases, RAM may be somewhat 44corrupt, but usually it is restorable. 45 462. Setting the parameters 47 48Setting the ramoops parameters can be done in 2 different manners: 49 1. Use the module parameters (which have the names of the variables described 50 as before). 51 For quick debugging, you can also reserve parts of memory during boot 52 and then use the reserved memory for ramoops. For example, assuming a machine 53 with > 128 MB of memory, the following kernel command line will tell the 54 kernel to use only the first 128 MB of memory, and place ECC-protected ramoops 55 region at 128 MB boundary: 56 "mem=128M ramoops.mem_address=0x8000000 ramoops.ecc=1" 57 2. Use a platform device and set the platform data. The parameters can then 58 be set through that platform data. An example of doing that is: 59 60#include <linux/pstore_ram.h> 61[...] 62 63static struct ramoops_platform_data ramoops_data = { 64 .mem_size = <...>, 65 .mem_address = <...>, 66 .mem_type = <...>, 67 .record_size = <...>, 68 .dump_oops = <...>, 69 .ecc = <...>, 70}; 71 72static struct platform_device ramoops_dev = { 73 .name = "ramoops", 74 .dev = { 75 .platform_data = &ramoops_data, 76 }, 77}; 78 79[... inside a function ...] 80int ret; 81 82ret = platform_device_register(&ramoops_dev); 83if (ret) { 84 printk(KERN_ERR "unable to register platform device\n"); 85 return ret; 86} 87 88You can specify either RAM memory or peripheral devices' memory. However, when 89specifying RAM, be sure to reserve the memory by issuing memblock_reserve() 90very early in the architecture code, e.g.: 91 92#include <linux/memblock.h> 93 94memblock_reserve(ramoops_data.mem_address, ramoops_data.mem_size); 95 963. Dump format 97 98The data dump begins with a header, currently defined as "====" followed by a 99timestamp and a new line. The dump then continues with the actual data. 100 1014. Reading the data 102 103The dump data can be read from the pstore filesystem. The format for these 104files is "dmesg-ramoops-N", where N is the record number in memory. To delete 105a stored record from RAM, simply unlink the respective pstore file. 106 1075. Persistent function tracing 108 109Persistent function tracing might be useful for debugging software or hardware 110related hangs. The functions call chain log is stored in a "ftrace-ramoops" 111file. Here is an example of usage: 112 113 # mount -t debugfs debugfs /sys/kernel/debug/ 114 # echo 1 > /sys/kernel/debug/pstore/record_ftrace 115 # reboot -f 116 [...] 117 # mount -t pstore pstore /mnt/ 118 # tail /mnt/ftrace-ramoops 119 0 ffffffff8101ea64 ffffffff8101bcda native_apic_mem_read <- disconnect_bsp_APIC+0x6a/0xc0 120 0 ffffffff8101ea44 ffffffff8101bcf6 native_apic_mem_write <- disconnect_bsp_APIC+0x86/0xc0 121 0 ffffffff81020084 ffffffff8101a4b5 hpet_disable <- native_machine_shutdown+0x75/0x90 122 0 ffffffff81005f94 ffffffff8101a4bb iommu_shutdown_noop <- native_machine_shutdown+0x7b/0x90 123 0 ffffffff8101a6a1 ffffffff8101a437 native_machine_emergency_restart <- native_machine_restart+0x37/0x40 124 0 ffffffff811f9876 ffffffff8101a73a acpi_reboot <- native_machine_emergency_restart+0xaa/0x1e0 125 0 ffffffff8101a514 ffffffff8101a772 mach_reboot_fixups <- native_machine_emergency_restart+0xe2/0x1e0 126 0 ffffffff811d9c54 ffffffff8101a7a0 __const_udelay <- native_machine_emergency_restart+0x110/0x1e0 127 0 ffffffff811d9c34 ffffffff811d9c80 __delay <- __const_udelay+0x30/0x40 128 0 ffffffff811d9d14 ffffffff811d9c3f delay_tsc <- __delay+0xf/0x20 129