blob: 2cb0a3080d795d7482e6b832438fcd55a7369c20 [file] [log] [blame]
Mike Frysinger74ce8322007-11-21 23:50:49 +08001menu "Kernel hacking"
2
3source "lib/Kconfig.debug"
4
Sonic Zhanga5ac0122008-10-13 14:07:19 +08005config HAVE_ARCH_KGDB
6 def_bool y
7
Mike Frysinger74ce8322007-11-21 23:50:49 +08008config DEBUG_MMRS
9 bool "Generate Blackfin MMR tree"
10 select DEBUG_FS
11 help
12 Create a tree of Blackfin MMRs via the debugfs tree. If
13 you enable this, you will find all MMRs laid out in the
14 /sys/kernel/debug/blackfin/ directory where you can read/write
15 MMRs directly from userspace. This is obviously just a debug
16 feature.
17
18config DEBUG_HWERR
19 bool "Hardware error interrupt debugging"
20 depends on DEBUG_KERNEL
21 help
22 When enabled, the hardware error interrupt is never disabled, and
23 will happen immediately when an error condition occurs. This comes
24 at a slight cost in code size, but is necessary if you are getting
25 hardware error interrupts and need to know where they are coming
26 from.
27
Robin Getz0c7a6b22008-10-08 16:27:12 +080028config DEBUG_DOUBLEFAULT
29 bool "Debug Double Faults"
30 default n
31 help
32 If an exception is caused while executing code within the exception
33 handler, the NMI handler, the reset vector, or in emulator mode,
34 a double fault occurs. On the Blackfin, this is a unrecoverable
35 event. You have two options:
36 - RESET exactly when double fault occurs. The excepting
37 instruction address is stored in RETX, where the next kernel
38 boot will print it out.
39 - Print debug message. This is much more error prone, although
40 easier to handle. It is error prone since:
41 - The excepting instruction is not committed.
42 - All writebacks from the instruction are prevented.
43 - The generated exception is not taken.
44 - The EXCAUSE field is updated with an unrecoverable event
45 The only way to check this is to see if EXCAUSE contains the
46 unrecoverable event value at every exception return. By selecting
47 this option, you are skipping over the faulting instruction, and
48 hoping things stay together enough to print out a debug message.
49
50 This does add a little kernel code, but is the only method to debug
51 double faults - if unsure say "Y"
52
53choice
54 prompt "Double Fault Failure Method"
55 default DEBUG_DOUBLEFAULT_PRINT
56 depends on DEBUG_DOUBLEFAULT
57
58config DEBUG_DOUBLEFAULT_PRINT
59 bool "Print"
60
61config DEBUG_DOUBLEFAULT_RESET
62 bool "Reset"
63
64endchoice
65
Mike Frysinger74ce8322007-11-21 23:50:49 +080066config DEBUG_ICACHE_CHECK
67 bool "Check Instruction cache coherency"
68 depends on DEBUG_KERNEL
69 depends on DEBUG_HWERR
70 help
71 Say Y here if you are getting weird unexplained errors. This will
72 ensure that icache is what SDRAM says it should be by doing a
73 byte wise comparison between SDRAM and instruction cache. This
74 also relocates the irq_panic() function to L1 memory, (which is
75 un-cached).
76
77config DEBUG_HUNT_FOR_ZERO
78 bool "Catch NULL pointer reads/writes"
79 default y
80 help
81 Say Y here to catch reads/writes to anywhere in the memory range
82 from 0x0000 - 0x0FFF (the first 4k) of memory. This is useful in
83 catching common programming errors such as NULL pointer dereferences.
84
85 Misbehaving applications will be killed (generate a SEGV) while the
86 kernel will trigger a panic.
87
88 Enabling this option will take up an extra entry in CPLB table.
89 Otherwise, there is no extra overhead.
90
91config DEBUG_BFIN_HWTRACE_ON
92 bool "Turn on Blackfin's Hardware Trace"
93 default y
94 help
95 All Blackfins include a Trace Unit which stores a history of the last
96 16 changes in program flow taken by the program sequencer. The history
97 allows the user to recreate the program sequencers recent path. This
98 can be handy when an application dies - we print out the execution
99 path of how it got to the offending instruction.
100
101 By turning this off, you may save a tiny amount of power.
102
103choice
104 prompt "Omit loop Tracing"
105 default DEBUG_BFIN_HWTRACE_COMPRESSION_OFF
106 depends on DEBUG_BFIN_HWTRACE_ON
107 help
108 The trace buffer can be configured to omit recording of changes in
109 program flow that match either the last entry or one of the last
110 two entries. Omitting one of these entries from the record prevents
111 the trace buffer from overflowing because of any sort of loop (for, do
112 while, etc) in the program.
113
114 Because zero-overhead Hardware loops are not recorded in the trace buffer,
115 this feature can be used to prevent trace overflow from loops that
116 are nested four deep.
117
118config DEBUG_BFIN_HWTRACE_COMPRESSION_OFF
119 bool "Trace all Loops"
120 help
121 The trace buffer records all changes of flow
122
123config DEBUG_BFIN_HWTRACE_COMPRESSION_ONE
124 bool "Compress single-level loops"
125 help
126 The trace buffer does not record single loops - helpful if trace
127 is spinning on a while or do loop.
128
129config DEBUG_BFIN_HWTRACE_COMPRESSION_TWO
130 bool "Compress two-level loops"
131 help
132 The trace buffer does not record loops two levels deep. Helpful if
133 the trace is spinning in a nested loop
134
135endchoice
136
137config DEBUG_BFIN_HWTRACE_COMPRESSION
138 int
139 depends on DEBUG_BFIN_HWTRACE_ON
140 default 0 if DEBUG_BFIN_HWTRACE_COMPRESSION_OFF
141 default 1 if DEBUG_BFIN_HWTRACE_COMPRESSION_ONE
142 default 2 if DEBUG_BFIN_HWTRACE_COMPRESSION_TWO
143
144
145config DEBUG_BFIN_HWTRACE_EXPAND
146 bool "Expand Trace Buffer greater than 16 entries"
147 depends on DEBUG_BFIN_HWTRACE_ON
148 default n
149 help
150 By selecting this option, every time the 16 hardware entries in
151 the Blackfin's HW Trace buffer are full, the kernel will move them
152 into a software buffer, for dumping when there is an issue. This
153 has a great impact on performance, (an interrupt every 16 change of
154 flows) and should normally be turned off, except in those nasty
155 debugging sessions
156
157config DEBUG_BFIN_HWTRACE_EXPAND_LEN
158 int "Size of Trace buffer (in power of 2k)"
159 range 0 4
160 depends on DEBUG_BFIN_HWTRACE_EXPAND
161 default 1
162 help
163 This sets the size of the software buffer that the trace information
164 is kept in.
165 0 for (2^0) 1k, or 256 entries,
166 1 for (2^1) 2k, or 512 entries,
167 2 for (2^2) 4k, or 1024 entries,
168 3 for (2^3) 8k, or 2048 entries,
169 4 for (2^4) 16k, or 4096 entries
170
171config DEBUG_BFIN_NO_KERN_HWTRACE
172 bool "Trace user apps (turn off hwtrace in kernel)"
173 depends on DEBUG_BFIN_HWTRACE_ON
174 default n
175 help
176 Some pieces of the kernel contain a lot of flow changes which can
177 quickly fill up the hardware trace buffer. When debugging crashes,
178 the hardware trace may indicate that the problem lies in kernel
179 space when in reality an application is buggy.
180
181 Say Y here to disable hardware tracing in some known "jumpy" pieces
182 of code so that the trace buffer will extend further back.
183
184config EARLY_PRINTK
185 bool "Early printk"
186 default n
187 help
188 This option enables special console drivers which allow the kernel
189 to print messages very early in the bootup process.
190
191 This is useful for kernel debugging when your machine crashes very
192 early before the console code is initialized. After enabling this
193 feature, you must add "earlyprintk=serial,uart0,57600" to the
194 command line (bootargs). It is safe to say Y here in all cases, as
195 all of this lives in the init section and is thrown away after the
196 kernel boots completely.
197
Mike Frysinger74ce8322007-11-21 23:50:49 +0800198config CPLB_INFO
199 bool "Display the CPLB information"
200 help
Mike Frysingercf934252008-02-02 15:32:40 +0800201 Display the CPLB information via /proc/cplbinfo.
Mike Frysinger74ce8322007-11-21 23:50:49 +0800202
203config ACCESS_CHECK
204 bool "Check the user pointer address"
205 default y
206 help
207 Usually the pointer transfer from user space is checked to see if its
208 address is in the kernel space.
209
210 Say N here to disable that check to improve the performance.
211
212endmenu