Commit | Line | Data |
---|---|---|
74ce8322 MF |
1 | menu "Kernel hacking" |
2 | ||
3 | source "lib/Kconfig.debug" | |
4 | ||
9f06c38f RG |
5 | config DEBUG_VERBOSE |
6 | bool "Verbose fault messages" | |
7 | default y | |
8 | select PRINTK | |
9 | help | |
10 | When a program crashes due to an exception, or the kernel detects | |
11 | an internal error, the kernel can print a not so brief message | |
12 | explaining what the problem was. This debugging information is | |
13 | useful to developers and kernel hackers when tracking down problems, | |
14 | but mostly meaningless to other people. This is always helpful for | |
15 | debugging but serves no purpose on a production system. | |
16 | Most people should say N here. | |
17 | ||
74ce8322 | 18 | config DEBUG_MMRS |
73ecfcf9 | 19 | tristate "Generate Blackfin MMR tree" |
74ce8322 MF |
20 | select DEBUG_FS |
21 | help | |
22 | Create a tree of Blackfin MMRs via the debugfs tree. If | |
23 | you enable this, you will find all MMRs laid out in the | |
24 | /sys/kernel/debug/blackfin/ directory where you can read/write | |
25 | MMRs directly from userspace. This is obviously just a debug | |
26 | feature. | |
27 | ||
28 | config DEBUG_HWERR | |
29 | bool "Hardware error interrupt debugging" | |
30 | depends on DEBUG_KERNEL | |
31 | help | |
32 | When enabled, the hardware error interrupt is never disabled, and | |
33 | will happen immediately when an error condition occurs. This comes | |
34 | at a slight cost in code size, but is necessary if you are getting | |
35 | hardware error interrupts and need to know where they are coming | |
36 | from. | |
37 | ||
b9a3899d RG |
38 | config EXACT_HWERR |
39 | bool "Try to make Hardware errors exact" | |
40 | depends on DEBUG_HWERR | |
41 | help | |
42 | By default, the Blackfin hardware errors are not exact - the error | |
43 | be reported multiple cycles after the error happens. This delay | |
44 | can cause the wrong application, or even the kernel to receive a | |
45 | signal to be killed. If you are getting HW errors in your system, | |
25985edc | 46 | try turning this on to ensure they are at least coming from the |
b9a3899d RG |
47 | proper thread. |
48 | ||
49 | On production systems, it is safe (and a small optimization) to say N. | |
50 | ||
0c7a6b21 RG |
51 | config DEBUG_DOUBLEFAULT |
52 | bool "Debug Double Faults" | |
53 | default n | |
54 | help | |
55 | If an exception is caused while executing code within the exception | |
56 | handler, the NMI handler, the reset vector, or in emulator mode, | |
57 | a double fault occurs. On the Blackfin, this is a unrecoverable | |
58 | event. You have two options: | |
59 | - RESET exactly when double fault occurs. The excepting | |
60 | instruction address is stored in RETX, where the next kernel | |
61 | boot will print it out. | |
62 | - Print debug message. This is much more error prone, although | |
63 | easier to handle. It is error prone since: | |
64 | - The excepting instruction is not committed. | |
65 | - All writebacks from the instruction are prevented. | |
66 | - The generated exception is not taken. | |
67 | - The EXCAUSE field is updated with an unrecoverable event | |
68 | The only way to check this is to see if EXCAUSE contains the | |
69 | unrecoverable event value at every exception return. By selecting | |
70 | this option, you are skipping over the faulting instruction, and | |
71 | hoping things stay together enough to print out a debug message. | |
72 | ||
73 | This does add a little kernel code, but is the only method to debug | |
74 | double faults - if unsure say "Y" | |
75 | ||
76 | choice | |
77 | prompt "Double Fault Failure Method" | |
78 | default DEBUG_DOUBLEFAULT_PRINT | |
79 | depends on DEBUG_DOUBLEFAULT | |
80 | ||
81 | config DEBUG_DOUBLEFAULT_PRINT | |
82 | bool "Print" | |
83 | ||
84 | config DEBUG_DOUBLEFAULT_RESET | |
85 | bool "Reset" | |
86 | ||
87 | endchoice | |
88 | ||
74ce8322 MF |
89 | config DEBUG_HUNT_FOR_ZERO |
90 | bool "Catch NULL pointer reads/writes" | |
91 | default y | |
92 | help | |
93 | Say Y here to catch reads/writes to anywhere in the memory range | |
94 | from 0x0000 - 0x0FFF (the first 4k) of memory. This is useful in | |
95 | catching common programming errors such as NULL pointer dereferences. | |
96 | ||
97 | Misbehaving applications will be killed (generate a SEGV) while the | |
98 | kernel will trigger a panic. | |
99 | ||
100 | Enabling this option will take up an extra entry in CPLB table. | |
101 | Otherwise, there is no extra overhead. | |
102 | ||
103 | config DEBUG_BFIN_HWTRACE_ON | |
104 | bool "Turn on Blackfin's Hardware Trace" | |
105 | default y | |
106 | help | |
107 | All Blackfins include a Trace Unit which stores a history of the last | |
108 | 16 changes in program flow taken by the program sequencer. The history | |
109 | allows the user to recreate the program sequencer’s recent path. This | |
110 | can be handy when an application dies - we print out the execution | |
111 | path of how it got to the offending instruction. | |
112 | ||
113 | By turning this off, you may save a tiny amount of power. | |
114 | ||
115 | choice | |
116 | prompt "Omit loop Tracing" | |
117 | default DEBUG_BFIN_HWTRACE_COMPRESSION_OFF | |
118 | depends on DEBUG_BFIN_HWTRACE_ON | |
119 | help | |
120 | The trace buffer can be configured to omit recording of changes in | |
121 | program flow that match either the last entry or one of the last | |
122 | two entries. Omitting one of these entries from the record prevents | |
123 | the trace buffer from overflowing because of any sort of loop (for, do | |
124 | while, etc) in the program. | |
125 | ||
126 | Because zero-overhead Hardware loops are not recorded in the trace buffer, | |
127 | this feature can be used to prevent trace overflow from loops that | |
128 | are nested four deep. | |
129 | ||
130 | config DEBUG_BFIN_HWTRACE_COMPRESSION_OFF | |
131 | bool "Trace all Loops" | |
132 | help | |
133 | The trace buffer records all changes of flow | |
134 | ||
135 | config DEBUG_BFIN_HWTRACE_COMPRESSION_ONE | |
136 | bool "Compress single-level loops" | |
137 | help | |
138 | The trace buffer does not record single loops - helpful if trace | |
139 | is spinning on a while or do loop. | |
140 | ||
141 | config DEBUG_BFIN_HWTRACE_COMPRESSION_TWO | |
142 | bool "Compress two-level loops" | |
143 | help | |
144 | The trace buffer does not record loops two levels deep. Helpful if | |
145 | the trace is spinning in a nested loop | |
146 | ||
147 | endchoice | |
148 | ||
149 | config DEBUG_BFIN_HWTRACE_COMPRESSION | |
150 | int | |
151 | depends on DEBUG_BFIN_HWTRACE_ON | |
152 | default 0 if DEBUG_BFIN_HWTRACE_COMPRESSION_OFF | |
153 | default 1 if DEBUG_BFIN_HWTRACE_COMPRESSION_ONE | |
154 | default 2 if DEBUG_BFIN_HWTRACE_COMPRESSION_TWO | |
155 | ||
156 | ||
157 | config DEBUG_BFIN_HWTRACE_EXPAND | |
158 | bool "Expand Trace Buffer greater than 16 entries" | |
159 | depends on DEBUG_BFIN_HWTRACE_ON | |
160 | default n | |
161 | help | |
162 | By selecting this option, every time the 16 hardware entries in | |
163 | the Blackfin's HW Trace buffer are full, the kernel will move them | |
164 | into a software buffer, for dumping when there is an issue. This | |
165 | has a great impact on performance, (an interrupt every 16 change of | |
166 | flows) and should normally be turned off, except in those nasty | |
167 | debugging sessions | |
168 | ||
169 | config DEBUG_BFIN_HWTRACE_EXPAND_LEN | |
170 | int "Size of Trace buffer (in power of 2k)" | |
171 | range 0 4 | |
172 | depends on DEBUG_BFIN_HWTRACE_EXPAND | |
173 | default 1 | |
174 | help | |
175 | This sets the size of the software buffer that the trace information | |
176 | is kept in. | |
177 | 0 for (2^0) 1k, or 256 entries, | |
178 | 1 for (2^1) 2k, or 512 entries, | |
179 | 2 for (2^2) 4k, or 1024 entries, | |
180 | 3 for (2^3) 8k, or 2048 entries, | |
181 | 4 for (2^4) 16k, or 4096 entries | |
182 | ||
183 | config DEBUG_BFIN_NO_KERN_HWTRACE | |
e60cb8a9 | 184 | bool "Turn off hwtrace in CPLB handlers" |
74ce8322 | 185 | depends on DEBUG_BFIN_HWTRACE_ON |
e60cb8a9 | 186 | default y |
74ce8322 | 187 | help |
e60cb8a9 | 188 | The CPLB error handler contains a lot of flow changes which can |
74ce8322 MF |
189 | quickly fill up the hardware trace buffer. When debugging crashes, |
190 | the hardware trace may indicate that the problem lies in kernel | |
191 | space when in reality an application is buggy. | |
192 | ||
193 | Say Y here to disable hardware tracing in some known "jumpy" pieces | |
194 | of code so that the trace buffer will extend further back. | |
195 | ||
196 | config EARLY_PRINTK | |
197 | bool "Early printk" | |
198 | default n | |
0cfbb323 | 199 | select SERIAL_CORE_CONSOLE |
74ce8322 MF |
200 | help |
201 | This option enables special console drivers which allow the kernel | |
202 | to print messages very early in the bootup process. | |
203 | ||
204 | This is useful for kernel debugging when your machine crashes very | |
205 | early before the console code is initialized. After enabling this | |
206 | feature, you must add "earlyprintk=serial,uart0,57600" to the | |
207 | command line (bootargs). It is safe to say Y here in all cases, as | |
208 | all of this lives in the init section and is thrown away after the | |
209 | kernel boots completely. | |
210 | ||
60ffdb36 GY |
211 | config NMI_WATCHDOG |
212 | bool "Enable NMI watchdog to help debugging lockup on SMP" | |
213 | default n | |
c429886a | 214 | depends on SMP |
60ffdb36 GY |
215 | help |
216 | If any CPU in the system does not execute the period local timer | |
217 | interrupt for more than 5 seconds, then the NMI handler dumps debug | |
218 | information. This information can be used to debug the lockup. | |
219 | ||
74ce8322 MF |
220 | config CPLB_INFO |
221 | bool "Display the CPLB information" | |
222 | help | |
cf93425d | 223 | Display the CPLB information via /proc/cplbinfo. |
74ce8322 MF |
224 | |
225 | config ACCESS_CHECK | |
226 | bool "Check the user pointer address" | |
227 | default y | |
228 | help | |
229 | Usually the pointer transfer from user space is checked to see if its | |
230 | address is in the kernel space. | |
231 | ||
232 | Say N here to disable that check to improve the performance. | |
233 | ||
c40cdb2c MF |
234 | config BFIN_ISRAM_SELF_TEST |
235 | bool "isram boot self tests" | |
236 | default n | |
237 | help | |
238 | Run some self tests of the isram driver code at boot. | |
239 | ||
6ce3e9c2 RG |
240 | config BFIN_PSEUDODBG_INSNS |
241 | bool "Support pseudo debug instructions" | |
242 | default n | |
243 | help | |
244 | This option allows the kernel to emulate some pseudo instructions which | |
245 | allow simulator test cases to be run under Linux with no changes. | |
246 | ||
247 | Most people should say N here. | |
248 | ||
72b099ed SZ |
249 | config BFIN_PM_WAKEUP_TIME_BENCH |
250 | bool "Display the total time for kernel to resume from power saving mode" | |
251 | default n | |
252 | help | |
253 | Display the total time when kernel resumes normal from standby or | |
254 | suspend to mem mode. | |
255 | ||
74ce8322 | 256 | endmenu |