This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH 3/3] Convert DTrace probe interface to C++ (and perform some cleanups)


On Tuesday, November 14 2017, Simon Marchi wrote:

> Hi Sergio,

Hey Simon,

Thanks for the review.

> On 2017-11-13 12:59 PM, Sergio Durigan Junior wrote:
>> This patch converts the DTrace probe
>> interface (gdb/dtrace-probe.[ch]) to C++, and also performs some
>> cleanups that were on my TODO list for a while.
>> 
>> The main changes were the conversion of 'struct dtrace_probe' to 'class
>> dtrace_probe', and a new 'class dtrace_static_probe_ops' to replace the
>> use of 'dtrace_probe_ops'.  Both classes implement the virtual methods
>> exported by their parents, 'class probe' and 'class static_probe_ops',
>> respectively.  I believe it's now a bit simpler to understand the
>> logic behind the dtrace-probe interface.
>> 
>> There are several helper functions used to parse parts of a dtrace
>> probe, and since they are generic and don't need to know about the
>> probe they're working on, I decided to leave them as simple static
>> functions (instead of e.g. converting them to class methods).
>> 
>> I've also converted a few uses of "VEC" to "std::vector", which makes
>> the code simpler and easier to maintain.  And, as usual, some cleanups
>> here and there.
>> 
>> Even though I'm sending a series of patches, they need to be tested
>> and committed as a single unit, because of inter-dependencies.  But it
>> should be easier to review in separate logical units.
>> 
>> I've regtested this patch on BuildBot, no regressions found.
>> 
>> gdb/ChangeLog:
>> 2017-11-13  Sergio Durigan Junior  <sergiodj@redhat.com>
>> 
>> 	* dtrace-probe.c (struct probe_ops dtrace_probe_ops): Delete.
>> 	(dtrace_probe_arg_s): Delete type and VEC.
>> 	(dtrace_probe_enabler_s): Likewise.
>> 	(struct dtrace_probe): Replace by...
>> 	(class dtrace_static_probe_ops): ...this and...
>> 	(class dtrace_probe): ...this.
>> 	(dtrace_probe_is_linespec): Rename to...
>> 	(dtrace_static_probe_ops::is_linespec): ...this.  Adjust code
>> 	to reflect change.
>> 	(dtrace_process_dof_probe): Use 'std::vector' instead of VEC.
>> 	Adjust code.  Create new instance of 'dtrace_probe'.
>> 	(dtrace_build_arg_exprs): Rename to...
>> 	(dtrace_probe::build_arg_exprs): ...this.  Adjust code to
>> 	reflect change.
>> 	(dtrace_get_probes): Rename to...
>> 	(dtrace_static_probe_ops::get_probes): ...this.  Adjust code
>> 	to reflect change.
>> 	(dtrace_get_arg): Rename to...
>> 	(dtrace_probe::get_arg_by_number): ...this.  Adjust code to
>> 	reflect change.
>> 	(dtrace_probe_is_enabled): Rename to...
>> 	(dtrace_probe::is_enabled): ...this.  Adjust code to reflect
>> 	change.
>> 	(dtrace_get_probe_address): Rename to...
>> 	(dtrace_probe::get_relocated_address): ...this.  Adjust code
>> 	to reflect change.
>> 	(dtrace_get_probe_argument_count): Rename to...
>> 	(dtrace_probe::get_argument_count): ...this.  Adjust code to
>> 	reflect change.
>> 	(dtrace_can_evaluate_probe_arguments): Rename to...
>> 	(dtrace_probe::can_evaluate_arguments): ...this.  Adjust code
>> 	to reflect change.
>> 	(dtrace_evaluate_probe_argument): Rename to...
>> 	(dtrace_probe::evaluate_argument): ...this.  Adjust code to
>> 	reflect change.
>> 	(dtrace_compile_to_ax): Rename to...
>> 	(dtrace_probe::compile_to_ax): ...this.  Adjust code to
>> 	reflect change.
>> 	(dtrace_probe_destroy): Delete.
>> 	(dtrace_type_name): Rename to...
>> 	(dtrace_static_probe_ops::type_name): ...this.  Adjust code to
>> 	reflect change.
>> 	(dtrace_probe::get_static_ops): New method.
>> 	(dtrace_gen_info_probes_table_header): Rename to...
>> 	(dtrace_static_probe_ops::gen_info_probes_table_header):
>> 	...this.  Adjust code to reflect change.
>> 	(dtrace_gen_info_probes_table_values): Rename to...
>> 	(dtrace_probe::gen_info_probes_table_values): ...this.  Adjust
>> 	code to reflect change.
>> 	(dtrace_enable_probe): Rename to...
>> 	(dtrace_probe::enable_probe): ...this.  Adjust code to reflect
>> 	change.
>> 	(dtrace_disable_probe): Rename to...
>> 	(dtrace_probe::disable_probe): ...this.  Adjust code to reflect
>> 	change.
>> 	(struct probe_ops dtrace_probe_ops): Delete.
>> 	(dtrace_static_probe_ops::emit_info_probes_extra_fields): New
>> 	method.
>> 	(info_probes_dtrace_command): Call 'info_probes_for_spops'
>> 	instead of 'info_probes_for_ops'.
>> 	(_initialize_dtrace_probe): Use 'all_static_probe_ops' instead
>> 	of 'all_probe_ops'.
>> ---
>>  gdb/dtrace-probe.c | 528 +++++++++++++++++++++++++++--------------------------
>>  1 file changed, 268 insertions(+), 260 deletions(-)
>> 
>> diff --git a/gdb/dtrace-probe.c b/gdb/dtrace-probe.c
>> index 2bbe03e4b3..b53ff7ac6c 100644
>> --- a/gdb/dtrace-probe.c
>> +++ b/gdb/dtrace-probe.c
>> @@ -41,10 +41,6 @@
>>  # define SHT_SUNW_dof	0x6ffffff4
>>  #endif
>>  
>> -/* Forward declaration.  */
>> -
>> -extern const struct probe_ops dtrace_probe_ops;
>> -
>>  /* The following structure represents a single argument for the
>>     probe.  */
>>  
>> @@ -60,9 +56,6 @@ struct dtrace_probe_arg
>>    struct expression *expr;
>>  };
>>  
>> -typedef struct dtrace_probe_arg dtrace_probe_arg_s;
>> -DEF_VEC_O (dtrace_probe_arg_s);
>> -
>>  /* The following structure represents an enabler for a probe.  */
>>  
>>  struct dtrace_probe_enabler
>> @@ -73,39 +66,121 @@ struct dtrace_probe_enabler
>>    CORE_ADDR address;
>>  };
>>  
>> -typedef struct dtrace_probe_enabler dtrace_probe_enabler_s;
>> -DEF_VEC_O (dtrace_probe_enabler_s);
>> +/* Class that implements the static probe methods for "stap" probes.  */
>> +
>> +class dtrace_static_probe_ops : public static_probe_ops
>> +{
>> +public:
>> +  /* See probe.h.  */
>> +  bool is_linespec (const char **linespecp) const override;
>> +
>> +  /* See probe.h.  */
>> +  void get_probes (std::vector<probe *> *probesp,
>> +		   struct objfile *objfile) const override;
>> +
>> +  /* See probe.h.  */
>> +  const char *type_name () const override;
>> +
>> +  /* See probe.h.  */
>> +  bool emit_info_probes_extra_fields () const override;
>> +
>> +  /* See probe.h.  */
>> +  void gen_info_probes_table_header
>> +    (std::vector<struct info_probe_column> *heads) const override;
>> +};
>> +
>> +/* DTrace static_probe_ops.  */
>> +
>> +const dtrace_static_probe_ops dtrace_static_probe_ops;
>>  
>>  /* The following structure represents a dtrace probe.  */
>>  
>> -struct dtrace_probe
>> +class dtrace_probe : public probe
>>  {
>> -  /* Generic information about the probe.  This must be the first
>> -     element of this struct, in order to maintain binary compatibility
>> -     with the `struct probe' and be able to fully abstract it.  */
>> -  struct probe p;
>> +public:
>> +  /* Constructor for dtrace_probe.  */
>> +  dtrace_probe (std::string &&name_, std::string &&provider_, CORE_ADDR address_,
>> +		struct gdbarch *arch_,
>> +		std::vector<struct dtrace_probe_arg> &args_,
>> +		std::vector<struct dtrace_probe_enabler> &enablers_)
>
> I am wondering why you don't use && for the vectors, as you do with the strings.
> If the constructor is "stealing" from the vectors passed as parameter, I think it
> would make sense if the parameters were &&, to indicate to the caller that they'll
> be moved from (and force it to use std::move).

That's true.  I think I was misunderstanding the relationship between
rvalues, lvalues and &&.  Fixed.

>> +    : probe (std::move (name_), std::move (provider_), address_, arch_),
>> +      m_args (std::move (args_)),
>> +      m_enablers (std::move (enablers_)),
>> +      m_args_expr_built (false)
>> +  {}
>> +
>> +  /* Destructor for dtrace_probe.  */
>> +  ~dtrace_probe ()
>> +  {
>> +    for (struct dtrace_probe_arg arg : m_args)
>> +      {
>> +	xfree (arg.type_str);
>> +	xfree (arg.expr);
>> +      }
>
> As for stap_probe, I think defining a destructor in dtrace_probe_arg would
> be clearer.

I used the same approach as in stap-probe.c: expression_up,
emplace_back, a constructor that takes care of std::move'ing things, and
no destructor.  I also converted 'type_str' to std::string.

>> +  }
>> +
>> +  /* See probe.h.  */
>> +  CORE_ADDR get_relocated_address (struct objfile *objfile) override;
>> +
>> +  /* See probe.h.  */
>> +  unsigned get_argument_count (struct frame_info *frame) override;
>> +
>> +  /* See probe.h.  */
>> +  int can_evaluate_arguments () const override;
>> +
>> +  /* See probe.h.  */
>> +  struct value *evaluate_argument (unsigned n,
>> +				   struct frame_info *frame) override;
>> +
>> +  /* See probe.h.  */
>> +  void compile_to_ax (struct agent_expr *aexpr,
>> +		      struct axs_value *axs_value,
>> +		      unsigned n) override;
>> +
>> +  /* See probe.h.  */
>> +  const static_probe_ops *get_static_ops () const override;
>> +
>> +  /* See probe.h.  */
>> +  void gen_info_probes_table_values
>> +    (std::vector<const char *> *values) const override;
>> +
>> +  /* See probe.h.  */
>> +  bool can_enable () const override
>> +  {
>> +    return true;
>> +  }
>> +
>> +  /* See probe.h.  */
>> +  void enable () override;
>>  
>> +  /* See probe.h.  */
>> +  void disable () override;
>> +
>> +  /* Return the Nth argument of the probe.  */
>> +  struct dtrace_probe_arg *get_arg_by_number (unsigned n,
>> +					      struct gdbarch *gdbarch);
>> +
>> +  /* Build the GDB internal expressiosn that, once evaluated, will
>> +     calculate the values of the arguments of the probe.  */
>> +  void build_arg_exprs (struct gdbarch *gdbarch);
>> +
>> +  /* Determine whether the probe is "enabled" or "disabled".  A
>> +     disabled probe is a probe in which one or more enablers are
>> +     disabled.  */
>> +  bool is_enabled () const;
>> +
>> +private:
>>    /* A probe can have zero or more arguments.  */
>> -  int probe_argc;
>> -  VEC (dtrace_probe_arg_s) *args;
>> +  int m_argc;
>
> It seems like this field is never set, but used at least once.  Should we use args->size() instead?

OK, no problem.  Done.

>> +  std::vector<struct dtrace_probe_arg> m_args;
>>  
>>    /* A probe can have zero or more "enablers" associated with it.  */
>> -  VEC (dtrace_probe_enabler_s) *enablers;
>> +  std::vector<struct dtrace_probe_enabler> m_enablers;
>>  
>>    /* Whether the expressions for the arguments have been built.  */
>> -  unsigned int args_expr_built : 1;
>> +  bool m_args_expr_built;
>>  };
>>  
>> -/* Implementation of the probe_is_linespec method.  */
>> -
>> -static int
>> -dtrace_probe_is_linespec (const char **linespecp)
>> -{
>> -  static const char *const keywords[] = { "-pdtrace", "-probe-dtrace", NULL };
>> -
>> -  return probe_is_linespec_by_keyword (linespecp, keywords);
>> -}
>> -
>>  /* DOF programs can contain an arbitrary number of sections of 26
>>     different types.  In order to support DTrace USDT probes we only
>>     need to handle a subset of these section types, fortunately.  These
>> @@ -322,8 +397,6 @@ dtrace_process_dof_probe (struct objfile *objfile,
>>  			  char *argtab, uint64_t strtab_size)
>>  {
>>    int i, j, num_probes, num_enablers;
>> -  struct cleanup *cleanup;
>> -  VEC (dtrace_probe_enabler_s) *enablers;
>>    char *p;
>>  
>>    /* Each probe section can define zero or more probes of two
>> @@ -368,9 +441,7 @@ dtrace_process_dof_probe (struct objfile *objfile,
>>  
>>    /* Build the list of enablers for the probes defined in this Probe
>>       DOF section.  */
>> -  enablers = NULL;
>> -  cleanup
>> -    = make_cleanup (VEC_cleanup (dtrace_probe_enabler_s), &enablers);
>> +  std::vector<struct dtrace_probe_enabler> enablers;
>>    num_enablers = DOF_UINT (dof, probe->dofpr_nenoffs);
>>    for (i = 0; i < num_enablers; i++)
>>      {
>> @@ -380,38 +451,32 @@ dtrace_process_dof_probe (struct objfile *objfile,
>>  
>>        enabler.address = DOF_UINT (dof, probe->dofpr_addr)
>>  	+ DOF_UINT (dof, enabler_offset);
>> -      VEC_safe_push (dtrace_probe_enabler_s, enablers, &enabler);
>> +      enablers.push_back (enabler);
>>      }
>>  
>>    for (i = 0; i < num_probes; i++)
>>      {
>>        uint32_t probe_offset
>>  	= ((uint32_t *) offtab)[DOF_UINT (dof, probe->dofpr_offidx) + i];
>> -      struct dtrace_probe *ret =
>> -	XOBNEW (&objfile->per_bfd->storage_obstack, struct dtrace_probe);
>> -
>> -      ret->p.pops = &dtrace_probe_ops;
>> -      ret->p.arch = gdbarch;
>> -      ret->args_expr_built = 0;
>>  
>>        /* Set the provider and the name of the probe.  */
>> -      ret->p.provider
>> -	= xstrdup (strtab + DOF_UINT (dof, provider->dofpv_name));
>> -      ret->p.name = xstrdup (strtab + DOF_UINT (dof, probe->dofpr_name));
>> +      const char *probe_provider
>> +	= strtab + DOF_UINT (dof, provider->dofpv_name);
>> +      const char *name = strtab + DOF_UINT (dof, probe->dofpr_name);
>>  
>>        /* The probe address.  */
>> -      ret->p.address
>> +      CORE_ADDR address
>>  	= DOF_UINT (dof, probe->dofpr_addr) + DOF_UINT (dof, probe_offset);
>>  
>>        /* Number of arguments in the probe.  */
>> -      ret->probe_argc = DOF_UINT (dof, probe->dofpr_nargc);
>> +      int probe_argc = DOF_UINT (dof, probe->dofpr_nargc);
>>  
>>        /* Store argument type descriptions.  A description of the type
>>           of the argument is in the (J+1)th null-terminated string
>>           starting at 'strtab' + 'probe->dofpr_nargv'.  */
>> -      ret->args = NULL;
>> +      std::vector<struct dtrace_probe_arg> args;
>>        p = strtab + DOF_UINT (dof, probe->dofpr_nargv);
>> -      for (j = 0; j < ret->probe_argc; j++)
>> +      for (j = 0; j < probe_argc; j++)
>>  	{
>>  	  struct dtrace_probe_arg arg;
>>  	  expression_up expr;
>> @@ -442,17 +507,18 @@ dtrace_process_dof_probe (struct objfile *objfile,
>>  	  if (expr != NULL && expr->elts[0].opcode == OP_TYPE)
>>  	    arg.type = expr->elts[1].type;
>>  
>> -	  VEC_safe_push (dtrace_probe_arg_s, ret->args, &arg);
>> +	  args.push_back (arg);
>>  	}
>>  
>> -      /* Add the vector of enablers to this probe, if any.  */
>> -      ret->enablers = VEC_copy (dtrace_probe_enabler_s, enablers);
>> +      std::vector<struct dtrace_probe_enabler> enablers_copy = enablers;
>
> Why is this copy necessary?  What happens with the original vector?

This copy is necessary because the original vector should be intact on
every iteration, but dtrace_probe's constructor performs a 'std::move'
on it.  I could have passed the original vector as an argument (by
value), and not perform the std::move in the constructor, if you prefer.

>> +      dtrace_probe *ret = new dtrace_probe (std::string (name),
>> +					    std::string (probe_provider),
>> +					    address, gdbarch,
>> +					    args, enablers_copy);
>>  
>>        /* Successfully created probe.  */
>> -      probesp->push_back ((struct probe *) ret);
>> +      probesp->push_back (ret);
>>      }
>> -
>> -  do_cleanups (cleanup);
>>  }
>>  
>>  /* Helper function to collect the probes described in the DOF program
>> @@ -555,28 +621,23 @@ dtrace_process_dof (asection *sect, struct objfile *objfile,
>>  	     sect->name);
>>  }
>>  
>> -/* Helper function to build the GDB internal expressiosn that, once
>> -   evaluated, will calculate the values of the arguments of a given
>> -   PROBE.  */
>> +/* Implementation of 'build_arg_exprs' method.  */
>>  
>> -static void
>> -dtrace_build_arg_exprs (struct dtrace_probe *probe,
>> -			struct gdbarch *gdbarch)
>> +void
>> +dtrace_probe::build_arg_exprs (struct gdbarch *gdbarch)
>>  {
>> -  struct parser_state pstate;
>> -  struct dtrace_probe_arg *arg;
>> -  int i;
>> -
>> -  probe->args_expr_built = 1;
>> +  m_args_expr_built = true;
>>  
>>    /* Iterate over the arguments in the probe and build the
>>       corresponding GDB internal expression that will generate the
>>       value of the argument when executed at the PC of the probe.  */
>> -  for (i = 0; i < probe->probe_argc; i++)
>> +  for (int i = 0; i < m_argc; i++)
>>      {
>> +      struct dtrace_probe_arg *arg;
>>        struct cleanup *back_to;
>> +      struct parser_state pstate;
>>  
>> -      arg = VEC_index (dtrace_probe_arg_s, probe->args, i);
>> +      arg = &m_args[i];
>>  
>>        /* Initialize the expression buffer in the parser state.  The
>>  	 language does not matter, since we are using our own
>> @@ -606,138 +667,82 @@ dtrace_build_arg_exprs (struct dtrace_probe *probe,
>>      }
>>  }
>>  
>> -/* Helper function to return the Nth argument of a given PROBE.  */
>> -
>> -static struct dtrace_probe_arg *
>> -dtrace_get_arg (struct dtrace_probe *probe, unsigned n,
>> -		struct gdbarch *gdbarch)
>> -{
>> -  if (!probe->args_expr_built)
>> -    dtrace_build_arg_exprs (probe, gdbarch);
>> -
>> -  return VEC_index (dtrace_probe_arg_s, probe->args, n);
>> -}
>> -
>> -/* Implementation of the get_probes method.  */
>> +/* Implementation of 'get_arg_by_number' method.  */
>>  
>> -static void
>> -dtrace_get_probes (std::vector<probe *> *probesp, struct objfile *objfile)
>> +struct dtrace_probe_arg *
>> +dtrace_probe::get_arg_by_number (unsigned n, struct gdbarch *gdbarch)
>>  {
>> -  bfd *abfd = objfile->obfd;
>> -  asection *sect = NULL;
>> -
>> -  /* Do nothing in case this is a .debug file, instead of the objfile
>> -     itself.  */
>> -  if (objfile->separate_debug_objfile_backlink != NULL)
>> -    return;
>> +  if (!m_args_expr_built)
>> +    this->build_arg_exprs (gdbarch);
>>  
>> -  /* Iterate over the sections in OBJFILE looking for DTrace
>> -     information.  */
>> -  for (sect = abfd->sections; sect != NULL; sect = sect->next)
>> -    {
>> -      if (elf_section_data (sect)->this_hdr.sh_type == SHT_SUNW_dof)
>> -	{
>> -	  bfd_byte *dof;
>> -
>> -	  /* Read the contents of the DOF section and then process it to
>> -	     extract the information of any probe defined into it.  */
>> -	  if (!bfd_malloc_and_get_section (abfd, sect, &dof))
>> -	    complaint (&symfile_complaints,
>> -		       _("could not obtain the contents of"
>> -			 "section '%s' in objfile `%s'."),
>> -		       sect->name, abfd->filename);
>> -      
>> -	  dtrace_process_dof (sect, objfile, probesp,
>> -			      (struct dtrace_dof_hdr *) dof);
>> -	  xfree (dof);
>> -	}
>> -    }
>> +  return &m_args[n];
>
> It would be nice to add a gdb_assert (n < m_args.size ()) here.

I've added a bound check and an internal_error, just like I did for
stap-probe.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]