Root/arch/blackfin/Kconfig.debug

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_HUNT_FOR_ZERO
106    bool "Catch NULL pointer reads/writes"
107    default y
108    help
109      Say Y here to catch reads/writes to anywhere in the memory range
110      from 0x0000 - 0x0FFF (the first 4k) of memory. This is useful in
111      catching common programming errors such as NULL pointer dereferences.
112
113      Misbehaving applications will be killed (generate a SEGV) while the
114      kernel will trigger a panic.
115
116      Enabling this option will take up an extra entry in CPLB table.
117      Otherwise, there is no extra overhead.
118
119config DEBUG_BFIN_HWTRACE_ON
120    bool "Turn on Blackfin's Hardware Trace"
121    default y
122    help
123      All Blackfins include a Trace Unit which stores a history of the last
124      16 changes in program flow taken by the program sequencer. The history
125      allows the user to recreate the program sequencer’s recent path. This
126      can be handy when an application dies - we print out the execution
127      path of how it got to the offending instruction.
128
129      By turning this off, you may save a tiny amount of power.
130
131choice
132    prompt "Omit loop Tracing"
133    default DEBUG_BFIN_HWTRACE_COMPRESSION_OFF
134    depends on DEBUG_BFIN_HWTRACE_ON
135    help
136      The trace buffer can be configured to omit recording of changes in
137      program flow that match either the last entry or one of the last
138      two entries. Omitting one of these entries from the record prevents
139      the trace buffer from overflowing because of any sort of loop (for, do
140      while, etc) in the program.
141
142      Because zero-overhead Hardware loops are not recorded in the trace buffer,
143      this feature can be used to prevent trace overflow from loops that
144      are nested four deep.
145
146config DEBUG_BFIN_HWTRACE_COMPRESSION_OFF
147    bool "Trace all Loops"
148    help
149      The trace buffer records all changes of flow
150
151config DEBUG_BFIN_HWTRACE_COMPRESSION_ONE
152    bool "Compress single-level loops"
153    help
154      The trace buffer does not record single loops - helpful if trace
155      is spinning on a while or do loop.
156
157config DEBUG_BFIN_HWTRACE_COMPRESSION_TWO
158    bool "Compress two-level loops"
159    help
160      The trace buffer does not record loops two levels deep. Helpful if
161      the trace is spinning in a nested loop
162
163endchoice
164
165config DEBUG_BFIN_HWTRACE_COMPRESSION
166    int
167    depends on DEBUG_BFIN_HWTRACE_ON
168    default 0 if DEBUG_BFIN_HWTRACE_COMPRESSION_OFF
169    default 1 if DEBUG_BFIN_HWTRACE_COMPRESSION_ONE
170    default 2 if DEBUG_BFIN_HWTRACE_COMPRESSION_TWO
171
172
173config DEBUG_BFIN_HWTRACE_EXPAND
174    bool "Expand Trace Buffer greater than 16 entries"
175    depends on DEBUG_BFIN_HWTRACE_ON
176    default n
177    help
178      By selecting this option, every time the 16 hardware entries in
179      the Blackfin's HW Trace buffer are full, the kernel will move them
180      into a software buffer, for dumping when there is an issue. This
181      has a great impact on performance, (an interrupt every 16 change of
182      flows) and should normally be turned off, except in those nasty
183      debugging sessions
184
185config DEBUG_BFIN_HWTRACE_EXPAND_LEN
186    int "Size of Trace buffer (in power of 2k)"
187    range 0 4
188    depends on DEBUG_BFIN_HWTRACE_EXPAND
189    default 1
190    help
191      This sets the size of the software buffer that the trace information
192      is kept in.
193      0 for (2^0) 1k, or 256 entries,
194      1 for (2^1) 2k, or 512 entries,
195      2 for (2^2) 4k, or 1024 entries,
196      3 for (2^3) 8k, or 2048 entries,
197      4 for (2^4) 16k, or 4096 entries
198
199config DEBUG_BFIN_NO_KERN_HWTRACE
200    bool "Turn off hwtrace in CPLB handlers"
201    depends on DEBUG_BFIN_HWTRACE_ON
202    default y
203    help
204      The CPLB error handler contains a lot of flow changes which can
205      quickly fill up the hardware trace buffer. When debugging crashes,
206      the hardware trace may indicate that the problem lies in kernel
207      space when in reality an application is buggy.
208
209      Say Y here to disable hardware tracing in some known "jumpy" pieces
210      of code so that the trace buffer will extend further back.
211
212config EARLY_PRINTK
213    bool "Early printk"
214    default n
215    select SERIAL_CORE_CONSOLE
216    help
217      This option enables special console drivers which allow the kernel
218      to print messages very early in the bootup process.
219
220      This is useful for kernel debugging when your machine crashes very
221      early before the console code is initialized. After enabling this
222      feature, you must add "earlyprintk=serial,uart0,57600" to the
223      command line (bootargs). It is safe to say Y here in all cases, as
224      all of this lives in the init section and is thrown away after the
225      kernel boots completely.
226
227config NMI_WATCHDOG
228    bool "Enable NMI watchdog to help debugging lockup on SMP"
229    default n
230    depends on SMP
231    help
232      If any CPU in the system does not execute the period local timer
233      interrupt for more than 5 seconds, then the NMI handler dumps debug
234      information. This information can be used to debug the lockup.
235
236config CPLB_INFO
237    bool "Display the CPLB information"
238    help
239      Display the CPLB information via /proc/cplbinfo.
240
241config ACCESS_CHECK
242    bool "Check the user pointer address"
243    default y
244    help
245      Usually the pointer transfer from user space is checked to see if its
246      address is in the kernel space.
247
248      Say N here to disable that check to improve the performance.
249
250config BFIN_ISRAM_SELF_TEST
251    bool "isram boot self tests"
252    default n
253    help
254      Run some self tests of the isram driver code at boot.
255
256config BFIN_PSEUDODBG_INSNS
257    bool "Support pseudo debug instructions"
258    default n
259    help
260      This option allows the kernel to emulate some pseudo instructions which
261      allow simulator test cases to be run under Linux with no changes.
262
263      Most people should say N here.
264
265endmenu
266

Archive Download this file



interactive