Save tracee memory map into trace.dat file
cover Save tracee memory map into trace.dat file
From: "Tzvetomir Stoyanov (VMware)" <[email protected]>
New trace-cmd option is introduced, "--mmap". When it is used together with -F or -P options, the memory map of the traced applications is stored into the trace.dat file. The new API tracecmd_search_tracee_mmap() can be used to do lookups into those memory maps.
Tzvetomir Stoyanov (1):
trace-cmd: Implemented new API tracecmd_add_option_v()
Tzvetomir Stoyanov (VMware) (1):
trace-cmd: Save the tracee memory map into the trace.dat file.
include/trace-cmd/trace-cmd.h | 9 +++ include/traceevent/event-parse.h | 1 + lib/trace-cmd/trace-input.c | 81 ++++++++++++++++++++ tracecmd/include/trace-local.h | 16 ++++ tracecmd/trace-output.c | 117 ++++++++++++++++++++++++----- tracecmd/trace-record.c | 123 +++++++++++++++++++++++++++++++ 6 files changed, 330 insertions(+), 17 deletions(-)
Signature · 1 line
2.21.0
PATCH 1/2 trace-cmd: Implemented new API tracecmd_add_option_v()
From: "Tzvetomir Stoyanov (VMware)" <[email protected]>
From: Tzvetomir Stoyanov <[email protected]>
This patch implements a new tracecmd API, tracecmd_add_option_v() It adds new option in trace.dat, similar to tracecmd_add_option(), but the option's data is passed as list of buffers. The standard struct iovec is used as input parameter, containing the option's data buffers.
--- include/trace-cmd/trace-cmd.h | 5 ++ include/traceevent/event-parse.h | 1 + tracecmd/trace-output.c | 117 ++++++++++++++++++++++++++----- 3 files changed, 106 insertions(+), 17 deletions(-)
include/trace-cmd/trace-cmd.h +5 -0
@@ -245,11 +245,16 @@ struct tracecmd_output *tracecmd_create_init_file_override(const char *output_fistruct tracecmd_option *tracecmd_add_option (struct tracecmd_output *handle,unsigned short id,int size,const void *data); +struct tracecmd_option * +tracecmd_add_option_v (struct tracecmd_output *handle, +unsigned short id,const struct iovec *vector,int count); +struct tracecmd_option *tracecmd_add_buffer_option (struct tracecmd_output *handle,const char *name,int cpus);int tracecmd_write_cpus (struct tracecmd_output *handle,int cpus);int tracecmd_write_options (struct tracecmd_output *handle); +int tracecmd_append_options (struct tracecmd_output *handle);int tracecmd_update_option (struct tracecmd_output *handle,struct tracecmd_option *option,int size,const void *data);
include/traceevent/event-parse.h +1 -0
@@ -11,6 +11,7 @@#include <stdio.h> #include <regex.h> #include <string.h> +#include <sys/uio.h> #include "trace-seq.h"
tracecmd/trace-output.c +100 -17
159 lines not on this page · raw message
Signature · 1 line
2.21.0
Re: [PATCH 1/2] trace-cmd: Implemented new API tracecmd_add_option_v()
Quoted 25 lines · On Mon, 17 Jun 2019 14:52:17 +0300 [email protected] wrote:
From: "Tzvetomir Stoyanov (VMware)" <[email protected]>
From: Tzvetomir Stoyanov <[email protected]>
This patch implements a new tracecmd API,
tracecmd_add_option_v()It adds new option in trace.dat, similar totracecmd_add_option(), but the option's data is passed as list of buffers. The standard struct iovec is used as input parameter, containing the option's data buffers.Signed-off-by Tzvetomir Stoyanov <[email protected]>--- include/trace-cmd/trace-cmd.h | 5 ++ include/traceevent/event-parse.h | 1 + tracecmd/trace-output.c | 117 ++++++++++++++++++++++++++----- 3 files changed, 106 insertions(+), 17 deletions(-)include/trace-cmd/trace-cmd.h +0
-0@@ -245,11 +245,16 @@ struct tracecmd_output *tracecmd_create_init_file_override(const char *output_fistruct tracecmd_option *tracecmd_add_option (struct tracecmd_output *handle,unsigned short id,int size,const void *data);
include/trace-cmd/trace-cmd.h struct tracecmd_output *tracecmd_create_init_file_override(const char *output_fi · line 257
Quoted 10 lines
include/trace-cmd/trace-cmd.h +5
-0+struct tracecmd_option * +tracecmd_add_option_v (struct tracecmd_output *handle, +unsigned short id,const struct iovec *vector,int count); +struct tracecmd_option *tracecmd_add_buffer_option (struct tracecmd_output *handle,const char *name,int cpus);int tracecmd_write_cpus (struct tracecmd_output *handle,int cpus);int tracecmd_write_options (struct tracecmd_output *handle); +int tracecmd_append_options (struct tracecmd_output *handle);
The change log doesn't mention anything about tracecmd_append_options()?
Quoted 31 lines
include/trace-cmd/trace-cmd.h +0
-0int tracecmd_update_option (struct tracecmd_output *handle,struct tracecmd_option *option,int size,const void *data);include/traceevent/event-parse.h +1
-0@@ -11,6 +11,7 @@#include <stdio.h> #include <regex.h> #include <string.h> +#include <sys/uio.h> #include "trace-seq.h" tracecmd/trace-output.c +3
-3@@ -883,21 +883,23 @@ static struct tracecmd_output *create_file(const char *output_file, }/** -* tracecmd_add_option - add options to the file +* tracecmd_add_option_v - add options to the file * @handle: the output file handle name * @id: the id of the option -* @size: the size of the option data -* @data: the data to write to the file. +* @vector: array of vectors, pointing to the data to write in the file +* @count: number of items in the vector array
tracecmd/trace-output.c static struct tracecmd_output *create_file(const char *output_file, · line 898
Quoted 10 lines
tracecmd/trace-output.c +2
-2* * Returns handle to update option if needed. * Just the content can be updated, with smaller or equal to * content than the specified size. */ struct tracecmd_option * -tracecmd_add_option (struct tracecmd_output *handle, -unsigned short id,int size,const void *data) +tracecmd_add_option_v (struct tracecmd_output *handle, +unsigned short id,const struct iovec *vector,int count)
Hmm, I think this is a bit overkill. I don't really see anything using more than one or two data vectors. All I see would be at most a "count" followed by a list of data, which is what I think you are using this for.
I rather wait to implement something like this when there's more of a need for it. I don't believe this change really requires it.
-- Steve
Quoted 140 lines
tracecmd/trace-output.c +95
-12{struct tracecmd_option *option; +char *data =NULL ; +int i, size =0 ;/* * We can only add options before they were written. @@ -906,32 +908,63 @@ tracecmd_add_option(struct tracecmd_output *handle,if (handle->options_written)return NULL ; - handle->nr_options++; +for (i =0 ; i < count; i++) + size += vector[i].iov_len; + +/* Some IDs (like TRACECMD_OPTION_TRACECLOCK) pass vector with 0 / NULL data */ +if (size) { + data =malloc (size); +if (!data) { +warning ("Insufficient memory" ); +return NULL ; + } + } option =malloc (sizeof (*option));if (!option) {warning ("Could not allocate space for option" ); +free (data);return NULL ; } - option->id = id; - option->size = size; - option->data =malloc (size); -if (!option->data) { -warning ("Insufficient memory" ); -free (option); -return NULL ; + handle->nr_options++; + option->data = data; +for (i =0 ; i < count; i++) { +if (vector[i].iov_base && vector[i].iov_len) { +memcpy (data, vector[i].iov_base, vector[i].iov_len); + data += vector[i].iov_len; + } } - -/* Some IDs (like TRACECMD_OPTION_TRACECLOCK) pass 0 / NULL data */ -if (size) -memcpy (option->data, data, size); + option->size = size; + option->id = id;list_add_tail (&option->list, &handle->options);return option; } +/** +* tracecmd_add_option - add options to the file +* @handle: the output file handle name +* @id: the id of the option +* @size: the size of the option data +* @data: the data to write to the file. +* +* Returns handle to update option if needed. +* Just the content can be updated, with smaller or equal to +* content than the specified size. +*/ +struct tracecmd_option * +tracecmd_add_option (struct tracecmd_output *handle, +unsigned short id,int size,const void *data) +{ +struct iovec vect; + + vect.iov_base = (void *) data; + vect.iov_len = size; +return tracecmd_add_option_v (handle, id, &vect,1 ); +} +int tracecmd_write_cpus (struct tracecmd_output *handle,int cpus) { cpus =convert_endian_4 (handle, cpus);@@ -979,6 +1012,56 @@ int tracecmd_write_options(struct tracecmd_output *handle)return 0 ; } +int tracecmd_append_options (struct tracecmd_output *handle) +{ +struct tracecmd_option *options; +unsigned short option; +unsigned short endian2; +unsigned int endian4; +off_t offset; +int r; + +/* If already written, ignore */ +if (handle->options_written) +return 0 ; + +if (lseek64 (handle->fd,0 ,SEEK_END ) == (off_t)-1 ) +return -1 ; + offset =lseek64 (handle->fd,-2 ,SEEK_CUR ); +if (offset == (off_t)-1 ) +return -1 ; + + r =pread (handle->fd, &option,2 , offset); +if (r !=2 || option !=TRACECMD_OPTION_DONE ) +return -1 ; + +list_for_each_entry (options, &handle->options, list) { + endian2 =convert_endian_2 (handle, options->id); +if (do_write_check (handle, &endian2,2 )) +return -1 ; + + endian4 =convert_endian_4 (handle, options->size); +if (do_write_check (handle, &endian4,4 )) +return -1 ; + +/* Save the data location in case it needs to be updated */ + options->offset =lseek64 (handle->fd,0 ,SEEK_CUR ); + +if (do_write_check (handle, options->data, + options->size)) +return -1 ; + } + + option =TRACECMD_OPTION_DONE ; + +if (do_write_check (handle, &option,2 )) +return -1 ; + + handle->options_written =1 ; + +return 0 ; +} +int tracecmd_update_option (struct tracecmd_output *handle,struct tracecmd_option *option,int size,const void *data)
Re: [PATCH 1/2] trace-cmd: Implemented new API tracecmd_add_option_v()
Hi Steven
Quoted 1 line · Steven Rostedt
...
Quoted 11 lines
Hmm, I think this is a bit overkill. I don't really see anything using more than one or two data vectors. All I see would be at most a "count" followed by a list of data, which is what I think you are using this for.
I rather wait to implement something like this when there's more of a need for it. I don't believe this change really requires it.
-- Steve
Actually, this patch is from the patch set (N 7):
"trace-cmd: Timetamps sync between host and guest machines, relying
on vsock events. "
and I took it as-is. In timesync changes tracecmd_add_option_v() is used in similar way.
That explains the tracecmd_append_options() API, which is used there.
Signature · 2 lines
Tzvetomir (Ceco) Stoyanov VMware Open Source Technology Center
Re: [PATCH 1/2] trace-cmd: Implemented new API tracecmd_add_option_v()
Quoted 17 lines · On Mon, 17 Jun 2019 16:18:38 +0300 Ceco <[email protected]> wrote:
Hi Steven
Quoted 1 line · Steven Rostedt
...
Quoted 11 lines
Hmm, I think this is a bit overkill. I don't really see anything using more than one or two data vectors. All I see would be at most a "count" followed by a list of data, which is what I think you are using this for.
I rather wait to implement something like this when there's more of a need for it. I don't believe this change really requires it.
-- Steve
Quoted 8 lines
Actually, this patch is from the patch set (N 7):
"trace-cmd: Timetamps sync between host and guest machines, relying
on vsock events. "
and I took it as-is. In timesync changestracecmd_add_option_v()is used in similar way.
That explains thetracecmd_append_options()API, which is used there.
Ah, that needed to be stated. Yeah, when taking a patch from another series, make sure that the change log is updated too. Never expect that the reviewer will know about any other series ;-)
-- Steve
PATCH 2/2 trace-cmd: Save the tracee memory map into the trace.dat file.
In reply to [email protected]From: "Tzvetomir Stoyanov (VMware)" <[email protected]>
A new trace-cmd record option is added: "--mmap". When it is set with combination of -F or -P options, the memory map of the traced applications is stored in the trace.dat file. A new API tracecmd_search_tracee_mmap() can be used to look up into stored memory maps. The map is retrieved from /proc/<pid>/maps file.
--- include/trace-cmd/trace-cmd.h | 4 ++ lib/trace-cmd/trace-input.c | 81 ++++++++++++++++++++++ tracecmd/include/trace-local.h | 16 +++++ tracecmd/trace-record.c | 123 +++++++++++++++++++++++++++++++++ 4 files changed, 224 insertions(+)
include/trace-cmd/trace-cmd.h +4 -0
@@ -81,6 +81,7 @@ enum {TRACECMD_OPTION_HOOK ,TRACECMD_OPTION_OFFSET ,TRACECMD_OPTION_CPUCOUNT , + TRACECMD_OPTION_PIDMMAPS, }; enum {
@@ -206,6 +207,9 @@ unsigned long long tracecmd_page_ts(struct tracecmd_input *handle,unsigned int tracecmd_record_ts_delta (struct tracecmd_input *handle,struct tep_record *record); +char *tracecmd_get_tracee_lib (struct tracecmd_input *handle, +int pid,unsigned long addr); +#ifndef SWIG /* hack for function graph work around */ extern __threadstruct tracecmd_input *tracecmd_curr_thread_handle;
lib/trace-cmd/trace-input.c +81 -0
99 lines not on this page · raw message
tracecmd/include/trace-local.h +16 -0
@@ -157,6 +157,20 @@ struct func_list {const char *mod; }; +struct lib_mem_map { +unsigned long long start; +unsigned long long end; +char *lib_name; +}; + +struct pid_mem_maps { +struct pid_mem_maps *next; +struct lib_mem_map *lib_maps; +unsigned int nr_lib_maps; +char *proc_name; +int pid; +}; +struct buffer_instance {struct buffer_instance *next;const char *name;
@@ -183,6 +197,8 @@ struct buffer_instance {struct tracecmd_msg_handle *msg_handle;struct tracecmd_output *network_handle; +struct pid_mem_maps *mem_maps; +int flags;int tracing_on_init_val;int tracing_on_fd;
tracecmd/trace-record.c +123 -0
183 lines not on this page · raw message
Signature · 1 line
2.21.0
Re: [PATCH 2/2] trace-cmd: Save the tracee memory map into the trace.dat file.
Quoted 51 lines · On Mon, 17 Jun 2019 14:52:18 +0300 [email protected] wrote:
From: "Tzvetomir Stoyanov (VMware)" <[email protected]>
A new trace-cmd record option is added: "--mmap". When it is set with combination of -F or -P options, the memory map of the traced applications is stored in the trace.dat file. A new API
tracecmd_search_tracee_mmap()can be used to look up into stored memory maps. The map is retrieved from /proc/<pid>/maps file.Signed-off-by Tzvetomir Stoyanov (VMware) <[email protected]>--- include/trace-cmd/trace-cmd.h | 4 ++ lib/trace-cmd/trace-input.c | 81 ++++++++++++++++++++++ tracecmd/include/trace-local.h | 16 +++++ tracecmd/trace-record.c | 123 +++++++++++++++++++++++++++++++++ 4 files changed, 224 insertions(+)include/trace-cmd/trace-cmd.h +4
-0@@ -81,6 +81,7 @@ enum {TRACECMD_OPTION_HOOK ,TRACECMD_OPTION_OFFSET ,TRACECMD_OPTION_CPUCOUNT , + TRACECMD_OPTION_PIDMMAPS, }; enum {@@ -206,6 +207,9 @@ unsigned long long tracecmd_page_ts(struct tracecmd_input *handle,unsigned int tracecmd_record_ts_delta (struct tracecmd_input *handle,struct tep_record *record); +char *tracecmd_get_tracee_lib (struct tracecmd_input *handle, +int pid,unsigned long addr); +#ifndef SWIG /* hack for function graph work around */ extern __threadstruct tracecmd_input *tracecmd_curr_thread_handle;lib/trace-cmd/trace-input.c +1
-0@@ -100,6 +100,7 @@ struct tracecmd_input {struct tracecmd_ftrace finfo;struct hook_list *hooks; +struct pid_mem_maps *pid_mmaps;/* file information */ size_t header_files_start;size_t ftrace_files_start;
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2137
Quoted 4 lines
lib/trace-cmd/trace-input.c +1
-0@@ -2133,6 +2134,83 @@ void tracecmd_set_ts2secs(struct tracecmd_input *handle, handle->use_trace_clock = false; } +static void trace_pid_mmap_load (struct tracecmd_input *handle,char *buf)
Probably should have a return value for this function.
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2146
Quoted 9 lines
lib/trace-cmd/trace-input.c +9
-0+{ +struct pid_mem_maps *maps; +char mapname[PATH_MAX +1 ]; +char *line; +int ret; +int i; + + maps =calloc (1 ,sizeof (*maps)); +if (!maps)
We can return -ENOMEM
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2156
Quoted 10 lines
lib/trace-cmd/trace-input.c +10
-0+return ; + + maps->nr_lib_maps =tep_read_number (handle->pevent, buf,4 ); + maps->lib_maps =calloc (maps->nr_lib_maps,sizeof (struct lib_mem_map )); +if (!maps->lib_maps) { +free (maps); +return ; + } + maps->next = handle->pid_mmaps; + handle->pid_mmaps = maps;
Probably want to hold off to adding maps till the end. That way we can check for more errors below.
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2158
Quoted 2 lines
lib/trace-cmd/trace-input.c +2
-0+ + ret =sscanf (buf+4 ,"%d %s" , &maps->pid, mapname);
Hmm, this is very dangerous. Using a %s here, where we don't know the length of that string can cause mapname to overflow. This is run as root, and I don't want someone to send a trace.dat file that can cause a security issue when read.
What you can do is something like this:
char tempbuf[PATH_MAX+1]; tempbuf[PATH_MAX] = '\0'; strncpy(tempbuf, buf+4, PATH_MAX); ret = sscanf(tempbuf, "%d %s", &maps->pid, mapname);
Which would make sure that buf wont overrun mapname.
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2160
Quoted 2 lines
lib/trace-cmd/trace-input.c +2
-0+if (ret ==2 ) + maps->proc_name =strdup (mapname);
Note, strdup() can fail. We need to check for that.
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2166
Quoted 6 lines
lib/trace-cmd/trace-input.c +6
-0+ line =strchr (buf+4 ,'\n' ); +for (i =0 ; i < maps->nr_lib_maps; i++) { +if (!line) +break ; + ret =sscanf (line+1 ,"%llx %llx %s" , &maps->lib_maps[i].start, + &maps->lib_maps[i].end, mapname);
We should make sure that line length is not greater than PATH_MAX and warn on it if it is. Otherwise it can overflow mapname.
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2168
Quoted 2 lines
lib/trace-cmd/trace-input.c +2
-0+if (ret ==3 ) + maps->lib_maps[i].lib_name =strdup (mapname);
Need to check the return of mapname.
Quoted 17 lines
lib/trace-cmd/trace-input.c +17
-0+ line =strchr (line+1 ,'\n' ); + } +} + +/** +* tracecmd_search_tracee_mmap - Search tracee memory address map +* @handle: input handle to the trace.dat file +* @pid: pid of the tracee +* @addr: address from the tracee memory space. +* +* Map of the tracee memory can be saved in the trace.dat file, using the option +* "--mmap". If there is such information, this API can be used to look up into +* this memory map to find library is loaded at the given @addr. +* +* The name of the library at given tracee @addr is returned. +*/ +char *tracecmd_search_tracee_mmap(struct tracecmd_input *handle,
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2195
Quoted 10 lines
lib/trace-cmd/trace-input.c +10
-0+ int pid, unsigned long long addr) +{ +struct pid_mem_maps *maps; +int i; + +if (!handle || !handle->pid_mmaps) +return NULL ; + + maps = handle->pid_mmaps; +while (maps) {
I wonder if we should sort the maps after they are loaded, and then we can do a binary search. Which would be much faster.
lib/trace-cmd/trace-input.c void tracecmd_set_ts2secs(struct tracecmd_input *handle, · line 2205
Quoted 10 lines
lib/trace-cmd/trace-input.c +10
-0+if (maps->pid == pid) +break ; + maps = maps->next; + } +if (!maps || !maps->nr_lib_maps || !maps->lib_maps) +return NULL ; + +for (i =0 ; i < maps->nr_lib_maps; i++) +if (maps->lib_maps[i].start <= addr && + maps->lib_maps[i].end > addr)
Probably could make this into a binary search too.
sort() and bsearch() are your friends ;-)
Quoted 26 lines
lib/trace-cmd/trace-input.c +11
-0+break ; + +if (i < maps->nr_lib_maps) +return maps->lib_maps[i].lib_name; + +return NULL ; +} +static int handle_options (struct tracecmd_input *handle) {unsigned long long offset;@@ -2229,6 +2307,9 @@ static int handle_options(struct tracecmd_input *handle) cpus = *(int *)buf; handle->cpus =tep_read_number (handle->pevent, &cpus,4 );break ; +case TRACECMD_OPTION_PIDMMAPS : +trace_pid_mmap_load (handle, buf); +break ;default :warning ("unknown option %d" , option);break ;tracecmd/include/trace-local.h +0
-0
tracecmd/include/trace-local.h struct func_list { · line 163
Quoted 7 lines
tracecmd/include/trace-local.h +4
-0@@ -157,6 +157,20 @@ struct func_list {const char *mod; }; +struct lib_mem_map { +unsigned long long start; +unsigned long long end; +char *lib_name;
To make this easier to read, lets add another tab between the type and name. Although, I haven't looked at this applied, so it may not be as bad as the patch looks.
Quoted 35 lines
tracecmd/include/trace-local.h +12
-0+}; + +struct pid_mem_maps { +struct pid_mem_maps *next; +struct lib_mem_map *lib_maps; +unsigned int nr_lib_maps; +char *proc_name; +int pid; +}; +struct buffer_instance {struct buffer_instance *next;const char *name;@@ -183,6 +197,8 @@ struct buffer_instance {struct tracecmd_msg_handle *msg_handle;struct tracecmd_output *network_handle; +struct pid_mem_maps *mem_maps; +int flags;int tracing_on_init_val;int tracing_on_fd;tracecmd/trace-record.c +1
-0@@ -83,6 +83,7 @@ static int max_kb;static bool use_tcp;static int do_ptrace; +static int get_mmap;static int filter_task;static int filter_pid =-1 ;
tracecmd/trace-record.c static char *make_pid_filter(char *curr_filter, const char *field) · line 1064
Quoted 4 lines
tracecmd/trace-record.c +1
-0@@ -1060,6 +1061,82 @@ static char *make_pid_filter(char *curr_filter, const char *field)return filter; } +static void get_pid_mmaps (int pid)
Although it's not tested, perhaps we should return an error if we come across one.
tracecmd/trace-record.c static char *make_pid_filter(char *curr_filter, const char *field) · line 1066
Quoted 2 lines
tracecmd/trace-record.c +2
-0+{ +char buf[PATH_MAX +100 ], perm[5 ], dev[6 ], mapname[PATH_MAX +1 ];
BTW, it's better not to put so many types on one line. It makes warning messages like "variable not used" easier to find. I personally prefer not to have any multiple declarations on a single line except for
counters: int i, j, k. All else really should be separate.
I know this is more of a personal preference, but I do think it makes the code easier to read.
tracecmd/trace-record.c static char *make_pid_filter(char *curr_filter, const char *field) · line 1069
Quoted 3 lines
tracecmd/trace-record.c +3
-0+struct buffer_instance *instance = &top_instance; +struct pid_mem_maps *maps = instance->mem_maps; +unsigned long long begin, end, inode, foo;
This isn't so bad to be on the same line, as they are related.
You could argue to say the char's above are related, but because they all have their own defined sizes, I find it to be a bit more messy.
Quoted 6 lines
tracecmd/trace-record.c +6
-0+struct lib_mem_map *map; +char fname[PATH_MAX +1 ]; +FILE *f; +int ret; +int i; +
tracecmd/trace-record.c static char *make_pid_filter(char *curr_filter, const char *field) · line 1085
Quoted 10 lines
tracecmd/trace-record.c +10
-0+while (maps) { +if (pid == maps->pid) +break ; + maps = maps->next; + } + +sprintf (fname,"/proc/%d/exe" , pid); + ret =readlink (fname, mapname,PATH_MAX ); +if (ret >=PATH_MAX || ret <0 ) +return ;
Probably should return error types, or at least do some kind of warning.
Quoted 13 lines
tracecmd/trace-record.c +13
-0+ +sprintf (fname,"/proc/%d/maps" , pid); + f =fopen (fname,"r" ); +if (!f) +return ; + +if (!maps) { + maps =calloc (1 ,sizeof (*maps)); +if (!maps) +return ; + maps->pid = pid; + maps->next = instance->mem_maps; + instance->mem_maps = maps;
tracecmd/trace-record.c static char *make_pid_filter(char *curr_filter, const char *field) · line 1108
Quoted 10 lines
tracecmd/trace-record.c +10
-0+ }else { +for (i =0 ; i < maps->nr_lib_maps; i++) +free (maps->lib_maps[i].lib_name); +free (maps->lib_maps); + maps->lib_maps =NULL ; + maps->nr_lib_maps =0 ; +free (maps->proc_name); + } + + maps->proc_name =strdup (mapname);
Check for strdup() return.
tracecmd/trace-record.c static char *make_pid_filter(char *curr_filter, const char *field) · line 1113
Quoted 5 lines
tracecmd/trace-record.c +5
-0+ +while (fgets (buf,sizeof (buf), f)) { + mapname[0 ] ='\0' ; + ret =sscanf (buf,"%llx-%llx %4s %llx %5s %lld %s" , + &begin, &end, perm, &foo, dev, &inode, mapname);
Probably should test the size of buf to make sure mapname isn't overflow.
Quoted 40 lines
tracecmd/trace-record.c +29
-0+if (ret ==7 && mapname[0 ] !='\0' ) { + map =realloc (maps->lib_maps, + (maps->nr_lib_maps+1 )*sizeof (*map)); +if (!map) +break ; + map[maps->nr_lib_maps].end = end; + map[maps->nr_lib_maps].start = begin; + map[maps->nr_lib_maps].lib_name =strdup (mapname); + maps->lib_maps = map; + maps->nr_lib_maps++; + } + } +fclose (f); +} + +static void get_filter_pid_mmaps (void ) +{ +struct filter_pids *p; + +for (p = filter_pids; p; p = p->next) { +if (p->exclude) +continue ; +get_pid_mmaps (p->pid); + } +} +static void update_task_filter (void ) {struct buffer_instance *instance;@@ -1068,6 +1145,9 @@ static void update_task_filter(void)if (no_filter)return ; +if (get_mmap && filter_pids) +get_filter_pid_mmaps (); +if (filter_task)add_filter_pid (pid,0 );
tracecmd/trace-record.c static void ptrace_wait(enum trace_type type, int main_pid) · line 1346
Quoted 5 lines
tracecmd/trace-record.c +2
-0@@ -1262,6 +1342,8 @@ static void ptrace_wait(enum trace_type type, int main_pid)break ;case PTRACE_EVENT_EXIT : +if (get_mmap) +get_pid_mmaps (main_pid);
I'll have to look at this a bit more. I want to make sure that we are not doing the full ptrace, and just tracing exit.
Quoted 21 lines
tracecmd/trace-record.c +14
-0ptrace (PTRACE_GETEVENTMSG , pid,NULL , &cstatus);ptrace (PTRACE_DETACH , pid,NULL ,NULL );break ;@@ -3092,6 +3174,35 @@ static void append_buffer(struct tracecmd_output *handle, } } +static void +add_pid_mem_maps (struct tracecmd_output *handle,struct buffer_instance *instance) +{ +struct pid_mem_maps *maps = instance->mem_maps; +struct iovec vector[2 ]; +struct trace_seq s; +int i; + +trace_seq_init (&s); +while (maps) { +if (!maps->nr_lib_maps) +continue ; +trace_seq_reset (&s); +trace_seq_printf (&s,"%d %s\n" , maps->pid, maps->proc_name);
tracecmd/trace-record.c static void append_buffer(struct tracecmd_output *handle, · line 3200
Quoted 10 lines
tracecmd/trace-record.c +10
-0+for (i =0 ; i < maps->nr_lib_maps; i++) +trace_seq_printf (&s,"%llx %llx %s\n" , + maps->lib_maps[i].start, + maps->lib_maps[i].end, + maps->lib_maps[i].lib_name); + vector[0 ].iov_len =4 ; + vector[0 ].iov_base = &maps->nr_lib_maps; + vector[1 ].iov_len = s.len+1 ; + vector[1 ].iov_base = s.buffer; +tracecmd_add_option_v (handle,TRACECMD_OPTION_PIDMMAPS , vector,2 );
As stated in the last email. The vector size is hard coded as 2, which is why I'm reluctant to add a vector version now. If it was dynamic, then that would be more of a reason. Here we can just create our own structure and use that:
struct tracecmd_map_file {
int nr_maps;
int maps_len;
char *maps;
}Or something like that.
-- Steve
Quoted 55 lines
tracecmd/trace-record.c +17
-0+ maps = maps->next; + } +trace_seq_destroy (&s); +} +static void add_buffer_stat (struct tracecmd_output *handle,struct buffer_instance *instance) {@@ -3272,6 +3383,10 @@ static void record_data(struct common_record_context *ctx)if (!no_top_instance () && !top_instance.msg_handle)print_stat (&top_instance); +for_all_instances (instance) { +add_pid_mem_maps (handle, instance); + } +tracecmd_append_cpu_data (handle, local_cpu_count, temp_files);for (i =0 ; i < max_cpu_count; i++)@@ -4382,6 +4497,7 @@ void update_first_instance(struct buffer_instance *instance, int topt) }enum { + OPT_mmap =244 , OPT_quiet =245 , OPT_debug =246 , OPT_no_filter =247 ,@@ -4612,6 +4728,7 @@ static void parse_record_options(int argc, {"debug", no_argument, NULL, OPT_debug}, {"quiet", no_argument, NULL, OPT_quiet}, {"help", no_argument, NULL, '?'}, + {"mmap", no_argument, NULL, OPT_mmap}, {"module", required_argument, NULL, OPT_module}, {NULL, 0, NULL, 0} };@@ -4843,6 +4960,9 @@ static void parse_record_options(int argc,case 'i' : ignore_event_not_found =1 ;break ; +case OPT_mmap: + get_mmap =1 ; +break ;case OPT_date: ctx->date =1 ;if (ctx->data_flags &DATA_FL_OFFSET )@@ -4909,6 +5029,9 @@ static void parse_record_options(int argc,add_func (&ctx->instance->filter_funcs, ctx->instance->filter_mod,"*" ); +if (filter_task && get_mmap) + do_ptrace =1 ; +if (do_ptrace && !filter_task && (filter_pid <0 ))die (" -c can only be used with -F (or -P with event-fork support)" );if (ctx->do_child && !filter_task &&! filter_pid)
Re: [PATCH 2/2] trace-cmd: Save the tracee memory map into the trace.dat file.
Quoted 5 lines · On Mon, 17 Jun 2019 09:37:22 -0400 Steven Rostedt <[email protected]> wrote:
As stated in the last email. The vector size is hard coded as 2, which is why I'm reluctant to add a vector version now. If it was dynamic, then that would be more of a reason. Here we can just create our own structure and use that:
Quoted 7 lines
struct tracecmd_map_file { int nr_maps; int maps_len; char *maps; }Or something like that.
As you can't send that to the tracecmd_add_option, you would need something more like this:
struct tracecmd_map_file {
int nr_maps;
int maps_len;
char maps[0];
};And then you would need to copy it first before sending it:
struct tracecmd_map_file *tmap; tmap = malloc(sizeof(*tmap) + s.len + 1); and copy the buffer: memcpy(&tmap->maps[0], s.buffer, s.len + 1);
before sending it to tracecmd_add_option();
-- Steve