[U-Boot] [RFC PATCH v1 1/2] kgdb: add breakpoint support

Peng Fan B51431 at freescale.com
Tue Nov 4 13:23:37 CET 2014


Hi Simon,

在 11/4/2014 2:46 PM, Simon Glass 写道:
> Hi Peng,
>
> On 31 October 2014 23:12, Peng Fan <Peng.Fan at freescale.com> wrote:
>> Add Z/z protocal support for breakpoint set/remove.
>>
>> Signed-off-by: Peng Fan <Peng.Fan at freescale.com>
>
> This looks good to me - I just have a few bits below.
>
>> ---
>>   common/kgdb.c  | 273 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>   include/kgdb.h |  35 ++++++++
>>   2 files changed, 308 insertions(+)
>>
>> diff --git a/common/kgdb.c b/common/kgdb.c
>> index d357463..fd83ccd 100644
>> --- a/common/kgdb.c
>> +++ b/common/kgdb.c
>> @@ -92,6 +92,8 @@
>>   #include <kgdb.h>
>>   #include <command.h>
>>
>> +#include <asm-generic/errno.h>
>
> #include <errno.h> would do
Ok.
>
>> +
>>   #undef KGDB_DEBUG
>
> Where is this used?
>
You mean KGDB_DEBUG?
>>
>>   /*
>> @@ -111,6 +113,17 @@ static int longjmp_on_fault = 0;
>>   static int kdebug = 1;
>>   #endif
>>
>> +struct kgdb_bkpt kgdb_break[KGDB_MAX_BREAKPOINTS] = {
>> +       [0 ... KGDB_MAX_BREAKPOINTS-1] = { .state = BP_UNDEFINED }
>> +};
>> +
>> +#ifdef CONFIG_ARM
>> +unsigned char gdb_bpt_instr[4] = {0xfe, 0xde, 0xff, 0xe7};
>> +#else
>> +#error "Please implement gdb_bpt_instr!"
>> +#endif
>> +
>> +
>>   static const char hexchars[]="0123456789abcdef";
>>
>>   /* Convert ch from a hex digit to an int */
>> @@ -309,6 +322,200 @@ putpacket(unsigned char *buffer)
>>          } while ((recv & 0x7f) != '+');
>>   }
>>
>> +int kgdb_validate_break_address(unsigned addr)
>> +{
>> +       /* Need More */
>
> ?
I'll remove the comment. Actually i just want to validate whether the 
add parameter is fine or not using this function.
>
>> +       return 0;
>> +}
>> +
>> +static int probe_kernel_read(unsigned char *dst, void *src, size_t size)
>> +{
>> +       int i;
>> +       unsigned char *dst_ptr = dst;
>> +       unsigned char *src_ptr = src;
>> +
>> +       for (i = 0; i < size; i++)
>> +               *dst_ptr++ = *src_ptr++;
>> +
>> +       return 0;
>> +}
>> +
>> +static int probe_kernel_write(char *dst, void *src, size_t size)
>
> These two above are strange function names - why 'kernel' - what does
> it mean in this context? Also could you must use memcpy(), either in
> the functions or instead of them?
Ok. memcpy is better. I just use the function name in linux kernel, 
maybe misleading here. how about probe_mem_read?
>
>> +{
>> +       int i;
>> +       char *dst_ptr = dst;
>> +       char *src_ptr = src;
>> +
>> +       for (i = 0; i < size; i++)
>> +               *dst_ptr++ = *src_ptr++;
>> +
>> +       return 0;
>> +}
>> +
>> +__weak int kgdb_arch_set_breakpoint(struct kgdb_bkpt *bpt)
>> +{
>> +       int err;
>> +
>> +       err = probe_kernel_read(bpt->saved_instr, (char *)bpt->bpt_addr,
>> +                               BREAK_INSTR_SIZE);
>> +       if (err)
>> +               return err;
>> +
>> +       err = probe_kernel_write((char *)bpt->bpt_addr, gdb_bpt_instr,
>> +                                BREAK_INSTR_SIZE);
>> +
>> +       return err;
>> +}
>> +
>> +/*
>> + * Set the breakpoints whose state is BP_SET to BP_ACTIVE
>> + */
>> +int kgdb_active_sw_breakpoints(void)
>> +{
>> +       int err;
>> +       int ret = 0;
>> +       int i;
>> +
>> +       for (i = 0; i < KGDB_MAX_BREAKPOINTS; i++) {
>> +               if (kgdb_break[i].state != BP_SET)
>> +                       continue;
>> +
>> +               err = kgdb_arch_set_breakpoint(&kgdb_break[i]);
>> +               if (err) {
>> +                       ret = err;
>> +                       printf("KGDB: BP install failed: %lx\n",
>> +                              kgdb_break[i].bpt_addr);
>> +                       continue;
>> +               }
>> +
>> +               kgdb_break[i].state = BP_ACTIVE;
>> +
>> +               /*
>> +                * kgdb_arch_set_breakpoint touched dcache and memory.
>> +                * cache should be flushed to let icache can see the updated
>> +                * inst.
>
> instruction
>
>> +                */
>> +               /* flush work is done in do_exit */
>> +               kgdb_flush_cache_all();
>> +       }
>> +
>> +       return ret;
>> +}
>> +
>> +/*
>> + * Set state from BP_SET to BP_REMOVED
>> + */
>> +int kgdb_remove_sw_breakpoint(unsigned int addr)
>> +{
>> +       int i;
>> +
>> +       for (i = 0; i < KGDB_MAX_BREAKPOINTS; i++) {
>> +               if ((kgdb_break[i].state == BP_SET) &&
>> +                   (kgdb_break[i].bpt_addr == addr)) {
>> +                       kgdb_break[i].state = BP_REMOVED;
>> +                       return 0;
>> +               }
>> +       }
>> +
>> +       return -ENOENT;
>> +}
>> +
>> +int kgdb_set_sw_breakpoint(unsigned int addr)
>> +{
>> +       int err = kgdb_validate_break_address(addr);
>> +       int breakno = -1;
>> +       int i;
>> +
>> +       if (err)
>> +               return err;
>> +
>> +       for (i = 0; i < KGDB_MAX_BREAKPOINTS; i++) {
>> +               if ((kgdb_break[i].state == BP_SET) &&
>> +                   (kgdb_break[i].bpt_addr == addr))
>> +                       return -EEXIST;
>> +       }
>> +
>> +       for (i = 0; i < KGDB_MAX_BREAKPOINTS; i++) {
>> +               /* removed or unused, use it */
>> +               if ((kgdb_break[i].state == BP_REMOVED) ||
>> +                   (kgdb_break[i].state == BP_UNDEFINED)) {
>> +                       breakno = i;
>> +                       break;
>> +               }
>> +       }
>> +
>> +       if (breakno == -1)
>> +               return -E2BIG;
>> +
>> +       kgdb_break[breakno].state = BP_SET;
>> +       kgdb_break[breakno].type = BP_BREAKPOINT;
>> +       kgdb_break[breakno].bpt_addr = addr;
>> +
>> +       return 0;
>> +}
>> +
>> +__weak int kgdb_arch_remove_breakpoint(struct kgdb_bkpt *bpt)
>> +{
>> +       return probe_kernel_write((char *)bpt->bpt_addr,
>> +                                 (char *)bpt->saved_instr, BREAK_INSTR_SIZE);
>> +}
>> +
>> +/*
>> + * set breakpoints whose state is BP_ACTIVE to BP_SET
>> + */
>> +int kgdb_deactivate_sw_breakpoints(void)
>> +{
>> +       int err;
>> +       int ret = 0;
>> +       int i;
>> +
>> +       for (i = 0; i < KGDB_MAX_BREAKPOINTS; i++) {
>> +               if (kgdb_break[i].state != BP_ACTIVE)
>> +                       continue;
>> +
>> +               err = kgdb_arch_remove_breakpoint(&kgdb_break[i]);
>> +               if (err) {
>> +                       printf("KGDB: BP remove failed: %lx\n",
>> +                              kgdb_break[i].bpt_addr);
>> +                       ret = err;
>> +               }
>> +
>> +               kgdb_break[i].state = BP_SET;
>> +
>> +               /*
>> +                * kgdb_arch_remove_breakpoint touched dcache and memory.
>> +                * cache should be flushed to let icache can see the updated
>> +                * inst.
>> +                */
>> +               kgdb_flush_cache_all();
>> +       }
>> +
>> +       return ret;
>> +}
>> +
>> +static int kgdb_remove_all_break(void)
>
> kgdb_remove_all_breakpoints() to be consistent?
kgdb_remove_all_breakpoints is better.
>
>> +{
>> +       int err;
>> +       int i;
>> +
>> +       /* Clear memory breakpoints. */
>> +       for (i = 0; i < KGDB_MAX_BREAKPOINTS; i++) {
>> +               if (kgdb_break[i].state != BP_ACTIVE)
>> +                       goto setundefined;
>> +               err = kgdb_arch_remove_breakpoint(&kgdb_break[i]);
>> +               if (err)
>> +                       printf("KGDB: breakpoint remove failed: %lx\n",
>> +                              kgdb_break[i].bpt_addr);
>> +setundefined:
>> +               kgdb_break[i].state = BP_UNDEFINED;
>> +       }
>> +
>> +       /* clear hardware breakpoints. */
>> +       /* ToDo in future. */
>
> /* TODO: clear hardware breakpoints. */
Ok.
>
>> +
>> +       return 0;
>> +}
>> +
>>   /*
>>    * This function does all command processing for interfacing to gdb.
>>    */
>> @@ -318,6 +525,7 @@ handle_exception (struct pt_regs *regs)
>>          int addr;
>>          int length;
>>          char *ptr;
>> +       char *bpt_type;
>>          kgdb_data kd;
>>          int i;
>>
>> @@ -376,6 +584,12 @@ handle_exception (struct pt_regs *regs)
>>
>>          putpacket((unsigned char *)&remcomOutBuffer);
>>
>> +       /*
>> +        * Each time trigger a kgdb break, first deactive all active
>> +        * breakpoints.
>> +        */
>> +       kgdb_deactivate_sw_breakpoints();
>> +
>>          while (1) {
>>                  volatile int errnum;
>>
>> @@ -394,6 +608,8 @@ handle_exception (struct pt_regs *regs)
>>                  if (errnum == 0) switch (remcomInBuffer[0]) {
>>
>>                  case '?':               /* report most recent signal */
>> +                       kgdb_remove_all_break();
>> +
>>                          remcomOutBuffer[0] = 'S';
>>                          remcomOutBuffer[1] = hexchars[kd.sigval >> 4];
>>                          remcomOutBuffer[2] = hexchars[kd.sigval & 0xf];
>> @@ -465,6 +681,8 @@ handle_exception (struct pt_regs *regs)
>>                                  kd.extype |= KGDBEXIT_WITHADDR;
>>                          }
>>
>> +                       kgdb_active_sw_breakpoints();
>> +
>>                          goto doexit;
>>
>>                  case 'S':    /* SSS  single step with signal SS */
>> @@ -479,6 +697,8 @@ handle_exception (struct pt_regs *regs)
>>                                  kd.extype |= KGDBEXIT_WITHADDR;
>>                          }
>>
>> +                       kgdb_active_sw_breakpoints();
>> +
>>                  doexit:
>>   /* Need to flush the instruction cache here, as we may have deposited a
>>    * breakpoint, and the icache probably has no way of knowing that a data ref to
>> @@ -506,6 +726,59 @@ handle_exception (struct pt_regs *regs)
>>                                  kgdb_error(KGDBERR_BADPARAMS);
>>                          }
>>                          break;
>> +               case 'z':
>> +                       /*
>> +                        * Break point remove
>> +                        * packet: zt,addr,length
>> +                        */
>> +               case 'Z':
>> +                       /*
>> +                        * Break point set
>> +                        * packet: Zt,addr,length
>> +                        *
>> +                        * t is type: `0' - software breakpoint,
>> +                        * `1' - hardware breakpoint, `2' - write watchpoint,
>> +                        * `3' - read watchpoint, `4' - access watchpoint;
>> +                        * addr is address; length is in bytes. For a software
>> +                        * breakpoint, length specifies the size of the
>> +                        * instruction to be patched. For hardware breakpoints
>> +                        * and watchpoints length specifies the memory region
>> +                        * to be monitored. To avoid potential problems with
>> +                        * duplicate packets, the operations should be
>> +                        * implemented in an idempotent way.
>> +                        */
>> +                       bpt_type = &remcomInBuffer[1];
>> +                       ptr = &remcomInBuffer[2];
>> +
>> +                       if (*(ptr++) != ',') {
>> +                               errnum = -EINVAL;
>> +                               break;
>> +                       }
>> +
>> +                       if (!hexToInt(&ptr, &addr)) {
>> +                               errnum = -EINVAL;
>> +                               break;
>> +                       }
>> +
>> +                       if (*ptr++ != ',') {
>> +                               errnum = -EINVAL;
>> +                               break;
>> +                       }
>> +
>> +                       /* only software breakpoint is implemented */
>> +                       if ((remcomInBuffer[0] == 'Z') && (*bpt_type == '0')) {
>> +                               errnum = kgdb_set_sw_breakpoint(addr);
>> +                       } else if ((remcomInBuffer[0] == 'z') &&
>> +                                (*bpt_type == '0')) {
>> +                               errnum = kgdb_remove_sw_breakpoint(addr);
>> +                       } else {
>> +                               /* Unsupported */
>> +                               errnum = -EINVAL;
>> +                       }
>> +
>> +                       if (errnum == 0)
>> +                               strcpy(remcomOutBuffer, "OK");
>> +                       break;
>>                  }                       /* switch */
>>
>>                  if (errnum != 0)
>> diff --git a/include/kgdb.h b/include/kgdb.h
>> index b6ba742..f9152b5 100644
>> --- a/include/kgdb.h
>> +++ b/include/kgdb.h
>> @@ -20,6 +20,12 @@
>>
>>   #define KGDBEXIT_WITHADDR      0x100
>>
>> +#if defined(CONFIG_ARM)
>> +#define BREAK_INSTR_SIZE       4
>> +#else
>> +#error "BREAK_INSTR_SIZE not set"
>> +#endif
>> +
>>   typedef
>>          struct {
>>                  int num;
>> @@ -38,6 +44,29 @@ typedef
>>          }
>>   kgdb_data;
>>
>> +enum kgdb_bptype {
>> +       BP_BREAKPOINT = 0,
>> +       BP_HARDWARE_BREAKPOINT,
>> +       BP_WRITE_WATCHPOINT,
>> +       BP_READ_WATCHPOINT,
>> +       BP_ACCESS_WATCHPOINT,
>> +       BP_POKE_BREAKPOINT,
>> +};
>> +
>> +enum kgdb_bpstate {
>> +       BP_UNDEFINED = 0,
>> +       BP_REMOVED,
>> +       BP_SET,
>> +       BP_ACTIVE
>> +};
>> +
>> +struct kgdb_bkpt {
>> +       unsigned long           bpt_addr;
>> +       unsigned char           saved_instr[BREAK_INSTR_SIZE];
>> +       enum kgdb_bptype        type;
>> +       enum kgdb_bpstate       state;
>> +};
>> +
>>   /* these functions are provided by the generic kgdb support */
>>   extern void kgdb_init(void);
>>   extern void kgdb_error(int);
>> @@ -67,4 +96,10 @@ extern void kgdb_interruptible(int);
>>   /* this is referenced in the trap handler for the platform */
>>   extern int (*debugger_exception_handler)(struct pt_regs *);
>>
>> +int kgdb_set_sw_break(unsigned int addr);
>> +int kgdb_remove_sw_break(unsigned int addr);
>> +int kgdb_validate_break_address(unsigned int addr);
>> +
>> +#define KGDB_MAX_BREAKPOINTS   32
>> +
>>   #endif /* __KGDB_H__ */
>> --
>> 1.8.4.5
>>
>
> Regards,
> Simon
>
Thanks for reviewing.

Regards,
Peng.


More information about the U-Boot mailing list