Mike Frysinger | 74ce832 | 2007-11-21 23:50:49 +0800 | [diff] [blame] | 1 | menu "Kernel hacking" |
| 2 | |
| 3 | source "lib/Kconfig.debug" |
| 4 | |
| 5 | config DEBUG_MMRS |
| 6 | bool "Generate Blackfin MMR tree" |
| 7 | select DEBUG_FS |
| 8 | help |
| 9 | Create a tree of Blackfin MMRs via the debugfs tree. If |
| 10 | you enable this, you will find all MMRs laid out in the |
| 11 | /sys/kernel/debug/blackfin/ directory where you can read/write |
| 12 | MMRs directly from userspace. This is obviously just a debug |
| 13 | feature. |
| 14 | |
| 15 | config DEBUG_HWERR |
| 16 | bool "Hardware error interrupt debugging" |
| 17 | depends on DEBUG_KERNEL |
| 18 | help |
| 19 | When enabled, the hardware error interrupt is never disabled, and |
| 20 | will happen immediately when an error condition occurs. This comes |
| 21 | at a slight cost in code size, but is necessary if you are getting |
| 22 | hardware error interrupts and need to know where they are coming |
| 23 | from. |
| 24 | |
| 25 | config DEBUG_ICACHE_CHECK |
| 26 | bool "Check Instruction cache coherency" |
| 27 | depends on DEBUG_KERNEL |
| 28 | depends on DEBUG_HWERR |
| 29 | help |
| 30 | Say Y here if you are getting weird unexplained errors. This will |
| 31 | ensure that icache is what SDRAM says it should be by doing a |
| 32 | byte wise comparison between SDRAM and instruction cache. This |
| 33 | also relocates the irq_panic() function to L1 memory, (which is |
| 34 | un-cached). |
| 35 | |
| 36 | config DEBUG_HUNT_FOR_ZERO |
| 37 | bool "Catch NULL pointer reads/writes" |
| 38 | default y |
| 39 | help |
| 40 | Say Y here to catch reads/writes to anywhere in the memory range |
| 41 | from 0x0000 - 0x0FFF (the first 4k) of memory. This is useful in |
| 42 | catching common programming errors such as NULL pointer dereferences. |
| 43 | |
| 44 | Misbehaving applications will be killed (generate a SEGV) while the |
| 45 | kernel will trigger a panic. |
| 46 | |
| 47 | Enabling this option will take up an extra entry in CPLB table. |
| 48 | Otherwise, there is no extra overhead. |
| 49 | |
| 50 | config DEBUG_BFIN_HWTRACE_ON |
| 51 | bool "Turn on Blackfin's Hardware Trace" |
| 52 | default y |
| 53 | help |
| 54 | All Blackfins include a Trace Unit which stores a history of the last |
| 55 | 16 changes in program flow taken by the program sequencer. The history |
| 56 | allows the user to recreate the program sequencer’s recent path. This |
| 57 | can be handy when an application dies - we print out the execution |
| 58 | path of how it got to the offending instruction. |
| 59 | |
| 60 | By turning this off, you may save a tiny amount of power. |
| 61 | |
| 62 | choice |
| 63 | prompt "Omit loop Tracing" |
| 64 | default DEBUG_BFIN_HWTRACE_COMPRESSION_OFF |
| 65 | depends on DEBUG_BFIN_HWTRACE_ON |
| 66 | help |
| 67 | The trace buffer can be configured to omit recording of changes in |
| 68 | program flow that match either the last entry or one of the last |
| 69 | two entries. Omitting one of these entries from the record prevents |
| 70 | the trace buffer from overflowing because of any sort of loop (for, do |
| 71 | while, etc) in the program. |
| 72 | |
| 73 | Because zero-overhead Hardware loops are not recorded in the trace buffer, |
| 74 | this feature can be used to prevent trace overflow from loops that |
| 75 | are nested four deep. |
| 76 | |
| 77 | config DEBUG_BFIN_HWTRACE_COMPRESSION_OFF |
| 78 | bool "Trace all Loops" |
| 79 | help |
| 80 | The trace buffer records all changes of flow |
| 81 | |
| 82 | config DEBUG_BFIN_HWTRACE_COMPRESSION_ONE |
| 83 | bool "Compress single-level loops" |
| 84 | help |
| 85 | The trace buffer does not record single loops - helpful if trace |
| 86 | is spinning on a while or do loop. |
| 87 | |
| 88 | config DEBUG_BFIN_HWTRACE_COMPRESSION_TWO |
| 89 | bool "Compress two-level loops" |
| 90 | help |
| 91 | The trace buffer does not record loops two levels deep. Helpful if |
| 92 | the trace is spinning in a nested loop |
| 93 | |
| 94 | endchoice |
| 95 | |
| 96 | config DEBUG_BFIN_HWTRACE_COMPRESSION |
| 97 | int |
| 98 | depends on DEBUG_BFIN_HWTRACE_ON |
| 99 | default 0 if DEBUG_BFIN_HWTRACE_COMPRESSION_OFF |
| 100 | default 1 if DEBUG_BFIN_HWTRACE_COMPRESSION_ONE |
| 101 | default 2 if DEBUG_BFIN_HWTRACE_COMPRESSION_TWO |
| 102 | |
| 103 | |
| 104 | config DEBUG_BFIN_HWTRACE_EXPAND |
| 105 | bool "Expand Trace Buffer greater than 16 entries" |
| 106 | depends on DEBUG_BFIN_HWTRACE_ON |
| 107 | default n |
| 108 | help |
| 109 | By selecting this option, every time the 16 hardware entries in |
| 110 | the Blackfin's HW Trace buffer are full, the kernel will move them |
| 111 | into a software buffer, for dumping when there is an issue. This |
| 112 | has a great impact on performance, (an interrupt every 16 change of |
| 113 | flows) and should normally be turned off, except in those nasty |
| 114 | debugging sessions |
| 115 | |
| 116 | config DEBUG_BFIN_HWTRACE_EXPAND_LEN |
| 117 | int "Size of Trace buffer (in power of 2k)" |
| 118 | range 0 4 |
| 119 | depends on DEBUG_BFIN_HWTRACE_EXPAND |
| 120 | default 1 |
| 121 | help |
| 122 | This sets the size of the software buffer that the trace information |
| 123 | is kept in. |
| 124 | 0 for (2^0) 1k, or 256 entries, |
| 125 | 1 for (2^1) 2k, or 512 entries, |
| 126 | 2 for (2^2) 4k, or 1024 entries, |
| 127 | 3 for (2^3) 8k, or 2048 entries, |
| 128 | 4 for (2^4) 16k, or 4096 entries |
| 129 | |
| 130 | config DEBUG_BFIN_NO_KERN_HWTRACE |
| 131 | bool "Trace user apps (turn off hwtrace in kernel)" |
| 132 | depends on DEBUG_BFIN_HWTRACE_ON |
| 133 | default n |
| 134 | help |
| 135 | Some pieces of the kernel contain a lot of flow changes which can |
| 136 | quickly fill up the hardware trace buffer. When debugging crashes, |
| 137 | the hardware trace may indicate that the problem lies in kernel |
| 138 | space when in reality an application is buggy. |
| 139 | |
| 140 | Say Y here to disable hardware tracing in some known "jumpy" pieces |
| 141 | of code so that the trace buffer will extend further back. |
| 142 | |
| 143 | config EARLY_PRINTK |
| 144 | bool "Early printk" |
| 145 | default n |
| 146 | help |
| 147 | This option enables special console drivers which allow the kernel |
| 148 | to print messages very early in the bootup process. |
| 149 | |
| 150 | This is useful for kernel debugging when your machine crashes very |
| 151 | early before the console code is initialized. After enabling this |
| 152 | feature, you must add "earlyprintk=serial,uart0,57600" to the |
| 153 | command line (bootargs). It is safe to say Y here in all cases, as |
| 154 | all of this lives in the init section and is thrown away after the |
| 155 | kernel boots completely. |
| 156 | |
| 157 | config DUAL_CORE_TEST_MODULE |
| 158 | tristate "Dual Core Test Module" |
| 159 | depends on (BF561) |
| 160 | default n |
| 161 | help |
| 162 | Say Y here to build-in dual core test module for dual core test. |
| 163 | |
| 164 | config CPLB_INFO |
| 165 | bool "Display the CPLB information" |
| 166 | help |
| 167 | Display the CPLB information. |
| 168 | |
| 169 | config ACCESS_CHECK |
| 170 | bool "Check the user pointer address" |
| 171 | default y |
| 172 | help |
| 173 | Usually the pointer transfer from user space is checked to see if its |
| 174 | address is in the kernel space. |
| 175 | |
| 176 | Say N here to disable that check to improve the performance. |
| 177 | |
| 178 | endmenu |