1menu "Kernel hacking" 2 3source "lib/Kconfig.debug" 4 5config DEBUG_STACKOVERFLOW 6 bool "Check for stack overflows" 7 depends on DEBUG_KERNEL 8 help 9 This option will cause messages to be printed if free stack space 10 drops below a certain limit. 11 12config DEBUG_STACK_USAGE 13 bool "Enable stack utilization instrumentation" 14 depends on DEBUG_KERNEL 15 help 16 Enables the display of the minimum amount of free stack which each 17 task has ever had available in the sysrq-T output. 18 19 This option will slow down process creation somewhat. 20 21config DEBUG_VERBOSE 22 bool "Verbose fault messages" 23 default y 24 select PRINTK 25 help 26 When a program crashes due to an exception, or the kernel detects 27 an internal error, the kernel can print a not so brief message 28 explaining what the problem was. This debugging information is 29 useful to developers and kernel hackers when tracking down problems, 30 but mostly meaningless to other people. This is always helpful for 31 debugging but serves no purpose on a production system. 32 Most people should say N here. 33 34config DEBUG_MMRS 35 bool "Generate Blackfin MMR tree" 36 select DEBUG_FS 37 help 38 Create a tree of Blackfin MMRs via the debugfs tree. If 39 you enable this, you will find all MMRs laid out in the 40 /sys/kernel/debug/blackfin/ directory where you can read/write 41 MMRs directly from userspace. This is obviously just a debug 42 feature. 43 44config DEBUG_HWERR 45 bool "Hardware error interrupt debugging" 46 depends on DEBUG_KERNEL 47 help 48 When enabled, the hardware error interrupt is never disabled, and 49 will happen immediately when an error condition occurs. This comes 50 at a slight cost in code size, but is necessary if you are getting 51 hardware error interrupts and need to know where they are coming 52 from. 53 54config EXACT_HWERR 55 bool "Try to make Hardware errors exact" 56 depends on DEBUG_HWERR 57 help 58 By default, the Blackfin hardware errors are not exact - the error 59 be reported multiple cycles after the error happens. This delay 60 can cause the wrong application, or even the kernel to receive a 61 signal to be killed. If you are getting HW errors in your system, 62 try turning this on to ensure they are at least comming from the 63 proper thread. 64 65 On production systems, it is safe (and a small optimization) to say N. 66 67config DEBUG_DOUBLEFAULT 68 bool "Debug Double Faults" 69 default n 70 help 71 If an exception is caused while executing code within the exception 72 handler, the NMI handler, the reset vector, or in emulator mode, 73 a double fault occurs. On the Blackfin, this is a unrecoverable 74 event. You have two options: 75 - RESET exactly when double fault occurs. The excepting 76 instruction address is stored in RETX, where the next kernel 77 boot will print it out. 78 - Print debug message. This is much more error prone, although 79 easier to handle. It is error prone since: 80 - The excepting instruction is not committed. 81 - All writebacks from the instruction are prevented. 82 - The generated exception is not taken. 83 - The EXCAUSE field is updated with an unrecoverable event 84 The only way to check this is to see if EXCAUSE contains the 85 unrecoverable event value at every exception return. By selecting 86 this option, you are skipping over the faulting instruction, and 87 hoping things stay together enough to print out a debug message. 88 89 This does add a little kernel code, but is the only method to debug 90 double faults - if unsure say "Y" 91 92choice 93 prompt "Double Fault Failure Method" 94 default DEBUG_DOUBLEFAULT_PRINT 95 depends on DEBUG_DOUBLEFAULT 96 97config DEBUG_DOUBLEFAULT_PRINT 98 bool "Print" 99 100config DEBUG_DOUBLEFAULT_RESET 101 bool "Reset" 102 103endchoice 104 105config DEBUG_ICACHE_CHECK 106 bool "Check Instruction cache coherency" 107 depends on DEBUG_KERNEL 108 depends on DEBUG_HWERR 109 help 110 Say Y here if you are getting weird unexplained errors. This will 111 ensure that icache is what SDRAM says it should be by doing a 112 byte wise comparison between SDRAM and instruction cache. This 113 also relocates the irq_panic() function to L1 memory, (which is 114 un-cached). 115 116config DEBUG_HUNT_FOR_ZERO 117 bool "Catch NULL pointer reads/writes" 118 default y 119 help 120 Say Y here to catch reads/writes to anywhere in the memory range 121 from 0x0000 - 0x0FFF (the first 4k) of memory. This is useful in 122 catching common programming errors such as NULL pointer dereferences. 123 124 Misbehaving applications will be killed (generate a SEGV) while the 125 kernel will trigger a panic. 126 127 Enabling this option will take up an extra entry in CPLB table. 128 Otherwise, there is no extra overhead. 129 130config DEBUG_BFIN_HWTRACE_ON 131 bool "Turn on Blackfin's Hardware Trace" 132 default y 133 help 134 All Blackfins include a Trace Unit which stores a history of the last 135 16 changes in program flow taken by the program sequencer. The history 136 allows the user to recreate the program sequencer���s recent path. This 137 can be handy when an application dies - we print out the execution 138 path of how it got to the offending instruction. 139 140 By turning this off, you may save a tiny amount of power. 141 142choice 143 prompt "Omit loop Tracing" 144 default DEBUG_BFIN_HWTRACE_COMPRESSION_OFF 145 depends on DEBUG_BFIN_HWTRACE_ON 146 help 147 The trace buffer can be configured to omit recording of changes in 148 program flow that match either the last entry or one of the last 149 two entries. Omitting one of these entries from the record prevents 150 the trace buffer from overflowing because of any sort of loop (for, do 151 while, etc) in the program. 152 153 Because zero-overhead Hardware loops are not recorded in the trace buffer, 154 this feature can be used to prevent trace overflow from loops that 155 are nested four deep. 156 157config DEBUG_BFIN_HWTRACE_COMPRESSION_OFF 158 bool "Trace all Loops" 159 help 160 The trace buffer records all changes of flow 161 162config DEBUG_BFIN_HWTRACE_COMPRESSION_ONE 163 bool "Compress single-level loops" 164 help 165 The trace buffer does not record single loops - helpful if trace 166 is spinning on a while or do loop. 167 168config DEBUG_BFIN_HWTRACE_COMPRESSION_TWO 169 bool "Compress two-level loops" 170 help 171 The trace buffer does not record loops two levels deep. Helpful if 172 the trace is spinning in a nested loop 173 174endchoice 175 176config DEBUG_BFIN_HWTRACE_COMPRESSION 177 int 178 depends on DEBUG_BFIN_HWTRACE_ON 179 default 0 if DEBUG_BFIN_HWTRACE_COMPRESSION_OFF 180 default 1 if DEBUG_BFIN_HWTRACE_COMPRESSION_ONE 181 default 2 if DEBUG_BFIN_HWTRACE_COMPRESSION_TWO 182 183 184config DEBUG_BFIN_HWTRACE_EXPAND 185 bool "Expand Trace Buffer greater than 16 entries" 186 depends on DEBUG_BFIN_HWTRACE_ON 187 default n 188 help 189 By selecting this option, every time the 16 hardware entries in 190 the Blackfin's HW Trace buffer are full, the kernel will move them 191 into a software buffer, for dumping when there is an issue. This 192 has a great impact on performance, (an interrupt every 16 change of 193 flows) and should normally be turned off, except in those nasty 194 debugging sessions 195 196config DEBUG_BFIN_HWTRACE_EXPAND_LEN 197 int "Size of Trace buffer (in power of 2k)" 198 range 0 4 199 depends on DEBUG_BFIN_HWTRACE_EXPAND 200 default 1 201 help 202 This sets the size of the software buffer that the trace information 203 is kept in. 204 0 for (2^0) 1k, or 256 entries, 205 1 for (2^1) 2k, or 512 entries, 206 2 for (2^2) 4k, or 1024 entries, 207 3 for (2^3) 8k, or 2048 entries, 208 4 for (2^4) 16k, or 4096 entries 209 210config DEBUG_BFIN_NO_KERN_HWTRACE 211 bool "Turn off hwtrace in CPLB handlers" 212 depends on DEBUG_BFIN_HWTRACE_ON 213 default y 214 help 215 The CPLB error handler contains a lot of flow changes which can 216 quickly fill up the hardware trace buffer. When debugging crashes, 217 the hardware trace may indicate that the problem lies in kernel 218 space when in reality an application is buggy. 219 220 Say Y here to disable hardware tracing in some known "jumpy" pieces 221 of code so that the trace buffer will extend further back. 222 223config EARLY_PRINTK 224 bool "Early printk" 225 default n 226 select SERIAL_CORE_CONSOLE 227 help 228 This option enables special console drivers which allow the kernel 229 to print messages very early in the bootup process. 230 231 This is useful for kernel debugging when your machine crashes very 232 early before the console code is initialized. After enabling this 233 feature, you must add "earlyprintk=serial,uart0,57600" to the 234 command line (bootargs). It is safe to say Y here in all cases, as 235 all of this lives in the init section and is thrown away after the 236 kernel boots completely. 237 238config NMI_WATCHDOG 239 bool "Enable NMI watchdog to help debugging lockup on SMP" 240 default n 241 depends on SMP 242 help 243 If any CPU in the system does not execute the period local timer 244 interrupt for more than 5 seconds, then the NMI handler dumps debug 245 information. This information can be used to debug the lockup. 246 247config CPLB_INFO 248 bool "Display the CPLB information" 249 help 250 Display the CPLB information via /proc/cplbinfo. 251 252config ACCESS_CHECK 253 bool "Check the user pointer address" 254 default y 255 help 256 Usually the pointer transfer from user space is checked to see if its 257 address is in the kernel space. 258 259 Say N here to disable that check to improve the performance. 260 261config BFIN_ISRAM_SELF_TEST 262 bool "isram boot self tests" 263 default n 264 help 265 Run some self tests of the isram driver code at boot. 266 267config BFIN_PSEUDODBG_INSNS 268 bool "Support pseudo debug instructions" 269 default n 270 help 271 This option allows the kernel to emulate some pseudo instructions which 272 allow simulator test cases to be run under Linux with no changes. 273 274 Most people should say N here. 275 276endmenu 277