tools/mm: Introduce a tool to handle entries in allocinfo
Hao Ge
reviewerPATCH 1/1 tools/mm: Introduce a tool to handle entries in allocinfo
From: Hao Ge <[email protected]>
Some users always say that the information provided by /proc/allocinfo is too extensive or bulky.
However, from allocinfo's own perspective, it is not at fault as it needs to provide comprehensive information. Users can then select the information that is useful to them based on their own needs.
For commonly used scenarios, we can develop a tool to facilitate users in filtering and utilizing relevant information from allocinfo.
Currently, there is only one filtering feature available, which is to filter out entries where the 'bytes' field is 0 (since it is often used to understand the current memory allocation status, as previously mentioned by David Wang, who noticed that 2/3 of the lines have an accumulative counter of 0, indicating no memory activities,it will fill up the entire screen,by using the "-s" parameter, a lot of unnecessary information for this scenario can be filtered out.)
In subsequent phases, we will continue to add more features to this tool, with the goal of making it convenient for most people to use the memory allocation profiling tool.
--- MAINTAINERS | 1 + tools/mm/Makefile | 4 +- tools/mm/allocinfo_tool.c | 150 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 153 insertions(+), 2 deletions(-) create mode 100644 tools/mm/allocinfo_tool.c
MAINTAINERS +1 -0
@@ -15000,6 +15000,7 @@ F: Documentation/mm/allocation-profiling.rst F: include/linux/alloc_tag.h F: include/linux/pgalloc_tag.h F: lib/alloc_tag.c +F: tools/mm/allocinfo_tool.c MEMORY CONTROLLER DRIVERS M: Krzysztof Kozlowski <[email protected]>
tools/mm/Makefile +2 -2
@@ -3,7 +3,7 @@ # include ../scripts/Makefile.include -BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test +BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool INSTALL_TARGETS = $(BUILD_TARGETS) thpmaps LIB_DIR = ../lib/api
@@ -23,7 +23,7 @@ $(LIBS): $(CC) $(CFLAGS) -o $@ $< $(LDFLAGS) clean: - $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test + $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool make -C $(LIB_DIR) clean sbindir ?= /usr/sbin
tools/mm/allocinfo_tool.c new file +150 -0
150 lines not on this page · raw message
Signature · 1 line
2.25.1
Suren Baghdasaryan
maintainerQuoted 21 lines · Hao Ge
From: Hao Ge <[email protected]>
Some users always say that the information provided by
/proc/allocinfois too extensive or bulky.However, from allocinfo's own perspective, it is not at fault as it needs to provide comprehensive information. Users can then select the information that is useful to them based on their own needs.
For commonly used scenarios, we can develop a tool to facilitate users in filtering and utilizing relevant information from allocinfo.
Currently, there is only one filtering feature available, which is to filter out entries where the 'bytes' field is 0 (since it is often used to understand the current memory allocation status, as previously mentioned by David Wang, who noticed that 2/3 of the lines have an accumulative counter of 0, indicating no memory activities,it will fill up the entire screen,by using the "-s" parameter, a lot of unnecessary information for this scenario can be filtered out.)
Quoted 3 lines
In subsequent phases, we will continue to add more features to this tool, with the goal of making it convenient for most people to use the memory allocation profiling tool.
CC'ing Alessio along with Pasha and Sourav who were interested in such a tool.
Hi Hao,
Thanks for the tool! Actually Alessio just developed a tool called alloctop (similar to slabtop) which I think will do what you want and more. It supports sorting, filtering, continuous update, etc. It's written in Rust and we are planning to upstream it once we finish testing and evaluating it on Android. Please take a look and see if it fits your usecase. Please also note that this tool has been implemented just last week, so hot off the press and might have some early bugs.
Thanks,
Suren.
Quoted 202 lines
Signed-off-by Hao Ge <[email protected]>--- MAINTAINERS | 1 + tools/mm/Makefile | 4 +- tools/mm/allocinfo_tool.c | 150 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 153 insertions(+), 2 deletions(-) create mode 100644 tools/mm/allocinfo_tool.cMAINTAINERS +1
-0@@ -15000,6 +15000,7 @@ F: Documentation/mm/allocation-profiling.rst F: include/linux/alloc_tag.h F: include/linux/pgalloc_tag.h F: lib/alloc_tag.c +F: tools/mm/allocinfo_tool.c MEMORY CONTROLLER DRIVERS M: Krzysztof Kozlowski <[email protected]>tools/mm/Makefile +2
-2@@ -3,7 +3,7 @@ # include ../scripts/Makefile.include -BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test +BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool INSTALL_TARGETS = $(BUILD_TARGETS) thpmaps LIB_DIR = ../lib/api@@ -23,7 +23,7 @@ $(LIBS): $(CC) $(CFLAGS) -o $@ $< $(LDFLAGS) clean: - $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test + $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool make -C $(LIB_DIR) clean sbindir ?= /usr/sbintools/mm/allocinfo_tool.c new file +150
-1@@ -0,0 +1,150 @@ +// SPDX-License-Identifier: GPL-2.0 +/* +* allocinfo_tool: Tool to parse allocinfo +* +* Authors: Hao Ge <[email protected]> +* +* Compile with: +* gcc -o allocinfo_tool allocinfo_tool.c +*/ + +#include <stdio.h> +#include <stdlib.h> +#include <sys/stat.h> +#include <string.h> +#include <getopt.h> + +#define ALLOCINFO_FILE "/proc/allocinfo" +#define BUF_SIZE 1024 +#define NAME_MAX_LENTH 64 + +struct alloc_tag_counters { +signed long long bytes; +unsigned long long calls; +}; + +struct codetag { +unsigned int lineno; +char modname[NAME_MAX_LENTH ]; +char function[NAME_MAX_LENTH ]; +char filename[NAME_MAX_LENTH ]; +}; + +struct alloc_info { +struct alloc_tag_counters counters; +struct codetag tag; +int has_modname; +}; + +static int arg_opt; + +enum OPT_BIT { +SKIP_ZERO =1 <<0 , +}; + +void usage (void ) +{ +printf ("Usage: ./allocinfo_tool [OPTIONS]\n" +"-s\t\t\tskip bytes for 0 allocinfo entries\n" + ); +} + +int parse_alloc_info (char *line,struct alloc_info *info) +{ +if (sscanf (line,"%12lli %8llu %[^:]:%d [%[^]]] func:%s" , + &info->counters.bytes, &info->counters.calls, + info->tag.filename, &info->tag.lineno, + info->tag.modname, info->tag.function) ==6 ){ + info->has_modname =1 ; +return 1 ; + }; + +if (sscanf (line,"%12llu %8llu %[^:]:%u func:%s" , + &info->counters.bytes, &info->counters.calls, + info->tag.filename, &info->tag.lineno, + info->tag.function) ==5 ){ + info->has_modname =0 ; +return 1 ; + } + +return 0 ; +} + +int read_alloc_info (void ) +{ +FILE *file =fopen (ALLOCINFO_FILE ,"r" ); + +if (!file) { +perror ("Failed to open /proc/allocinfo" ); +return EXIT_FAILURE ; + } + +int line =0 , i =0 ; +char *buffer =malloc (BUF_SIZE ); +struct alloc_info *info; + +while (fgets (buffer,BUF_SIZE , file)) { + +/* +* allocinfo - version: 1.0 +* # <size> <calls> <tag info> +*/ +if (line <2 ) { +printf ("%s" , buffer); + line++; +continue ; + } + + info =realloc (info,sizeof (struct alloc_info ) * (i +1 )); + +if (parse_alloc_info (buffer, info + i) ==0 ) { +printf ("Mismatch with the format of /proc/allocinfo" ); +return 0 ; + } + +if ((arg_opt &SKIP_ZERO ) && (info[i].counters.bytes ==0 )) +continue ; + +printf ("%12lli %8llu " , info[i].counters.bytes, info[i].counters.calls); + +if (info[i].has_modname) +printf ("%s:%u [%s] func:%s" , + info[i].tag.filename, info[i].tag.lineno, + info[i].tag.modname, info[i].tag.function); +else +printf ("%s:%u func:%s" , + info[i].tag.filename, info[i].tag.lineno, + info[i].tag.function); +printf (" " ); +printf ("\n" ); + i++; + } + +free (info); +free (buffer); +fclose (file); +} + +int main (int argc,char *argv[]) +{ + +int opt; +struct option longopts[] = { + {"s" ,0 ,NULL ,1 }, + {0 ,0 ,0 ,0 }, + }; + +while ((opt =getopt_long (argc, argv,"s" , longopts,NULL )) !=-1 ) { +switch (opt) { +case 's' : + arg_opt = arg_opt |SKIP_ZERO ; +break ; +default : +usage (); +exit (1 ); + } + } + +read_alloc_info (); + +} --2.25.1
Thank you Suren for looping me in and Hao for the tool!
As Suren mentioned, we recently developed alloctop for the same reason as
yours: easy and fast manipulation of /proc/allocinfo instead of having long chains of piped sorts, greps, awks...alloctop takes slabtop as a reference for command line parameters and output format.
It currently supports sorting and filtering by minimum allocation size, as well as some kind of aggregation based on tag (similar to what du does). As Suren mentioned, the tool is currently being tested on Android, but we would be happy to collaborate on adding new features and making it part of tools/mm.
Thanks,
Alessio
Quoted 246 lines · Suren Baghdasaryan
Quoted 21 lines · Hao Ge
From: Hao Ge <[email protected]>
Some users always say that the information provided by
/proc/allocinfois too extensive or bulky.However, from allocinfo's own perspective, it is not at fault as it needs to provide comprehensive information. Users can then select the information that is useful to them based on their own needs.
For commonly used scenarios, we can develop a tool to facilitate users in filtering and utilizing relevant information from allocinfo.
Currently, there is only one filtering feature available, which is to filter out entries where the 'bytes' field is 0 (since it is often used to understand the current memory allocation status, as previously mentioned by David Wang, who noticed that 2/3 of the lines have an accumulative counter of 0, indicating no memory activities,it will fill up the entire screen,by using the "-s" parameter, a lot of unnecessary information for this scenario can be filtered out.)
Quoted 3 lines
In subsequent phases, we will continue to add more features to this tool, with the goal of making it convenient for most people to use the memory allocation profiling tool.
CC'ing Alessio along with Pasha and Sourav who were interested in such a tool.
Hi Hao,
Thanks for the tool! Actually Alessio just developed a tool called alloctop (similar to slabtop) which I think will do what you want and more. It supports sorting, filtering, continuous update, etc. It's written in Rust and we are planning to upstream it once we finish testing and evaluating it on Android. Please take a look and see if it fits your usecase. Please also note that this tool has been implemented just last week, so hot off the press and might have some early bugs.
Thanks,
Suren.Quoted 202 lines
Signed-off-by Hao Ge <[email protected]>--- MAINTAINERS | 1 + tools/mm/Makefile | 4 +- tools/mm/allocinfo_tool.c | 150 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 153 insertions(+), 2 deletions(-) create mode 100644 tools/mm/allocinfo_tool.cMAINTAINERS +1
-0@@ -15000,6 +15000,7 @@ F: Documentation/mm/allocation-profiling.rst F: include/linux/alloc_tag.h F: include/linux/pgalloc_tag.h F: lib/alloc_tag.c +F: tools/mm/allocinfo_tool.c MEMORY CONTROLLER DRIVERS M: Krzysztof Kozlowski <[email protected]>tools/mm/Makefile +2
-2@@ -3,7 +3,7 @@ # include ../scripts/Makefile.include -BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test +BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool INSTALL_TARGETS = $(BUILD_TARGETS) thpmaps LIB_DIR = ../lib/api@@ -23,7 +23,7 @@ $(LIBS): $(CC) $(CFLAGS) -o $@ $< $(LDFLAGS) clean: - $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test + $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool make -C $(LIB_DIR) clean sbindir ?= /usr/sbintools/mm/allocinfo_tool.c new file +150
-1@@ -0,0 +1,150 @@ +// SPDX-License-Identifier: GPL-2.0 +/* +* allocinfo_tool: Tool to parse allocinfo +* +* Authors: Hao Ge <[email protected]> +* +* Compile with: +* gcc -o allocinfo_tool allocinfo_tool.c +*/ + +#include <stdio.h> +#include <stdlib.h> +#include <sys/stat.h> +#include <string.h> +#include <getopt.h> + +#define ALLOCINFO_FILE "/proc/allocinfo" +#define BUF_SIZE 1024 +#define NAME_MAX_LENTH 64 + +struct alloc_tag_counters { +signed long long bytes; +unsigned long long calls; +}; + +struct codetag { +unsigned int lineno; +char modname[NAME_MAX_LENTH ]; +char function[NAME_MAX_LENTH ]; +char filename[NAME_MAX_LENTH ]; +}; + +struct alloc_info { +struct alloc_tag_counters counters; +struct codetag tag; +int has_modname; +}; + +static int arg_opt; + +enum OPT_BIT { +SKIP_ZERO =1 <<0 , +}; + +void usage (void ) +{ +printf ("Usage: ./allocinfo_tool [OPTIONS]\n" +"-s\t\t\tskip bytes for 0 allocinfo entries\n" + ); +} + +int parse_alloc_info (char *line,struct alloc_info *info) +{ +if (sscanf (line,"%12lli %8llu %[^:]:%d [%[^]]] func:%s" , + &info->counters.bytes, &info->counters.calls, + info->tag.filename, &info->tag.lineno, + info->tag.modname, info->tag.function) ==6 ){ + info->has_modname =1 ; +return 1 ; + }; + +if (sscanf (line,"%12llu %8llu %[^:]:%u func:%s" , + &info->counters.bytes, &info->counters.calls, + info->tag.filename, &info->tag.lineno, + info->tag.function) ==5 ){ + info->has_modname =0 ; +return 1 ; + } + +return 0 ; +} + +int read_alloc_info (void ) +{ +FILE *file =fopen (ALLOCINFO_FILE ,"r" ); + +if (!file) { +perror ("Failed to open /proc/allocinfo" ); +return EXIT_FAILURE ; + } + +int line =0 , i =0 ; +char *buffer =malloc (BUF_SIZE ); +struct alloc_info *info; + +while (fgets (buffer,BUF_SIZE , file)) { + +/* +* allocinfo - version: 1.0 +* # <size> <calls> <tag info> +*/ +if (line <2 ) { +printf ("%s" , buffer); + line++; +continue ; + } + + info =realloc (info,sizeof (struct alloc_info ) * (i +1 )); + +if (parse_alloc_info (buffer, info + i) ==0 ) { +printf ("Mismatch with the format of /proc/allocinfo" ); +return 0 ; + } + +if ((arg_opt &SKIP_ZERO ) && (info[i].counters.bytes ==0 )) +continue ; + +printf ("%12lli %8llu " , info[i].counters.bytes, info[i].counters.calls); + +if (info[i].has_modname) +printf ("%s:%u [%s] func:%s" , + info[i].tag.filename, info[i].tag.lineno, + info[i].tag.modname, info[i].tag.function); +else +printf ("%s:%u func:%s" , + info[i].tag.filename, info[i].tag.lineno, + info[i].tag.function); +printf (" " ); +printf ("\n" ); + i++; + } + +free (info); +free (buffer); +fclose (file); +} + +int main (int argc,char *argv[]) +{ + +int opt; +struct option longopts[] = { + {"s" ,0 ,NULL ,1 }, + {0 ,0 ,0 ,0 }, + }; + +while ((opt =getopt_long (argc, argv,"s" , longopts,NULL )) !=-1 ) { +switch (opt) { +case 's' : + arg_opt = arg_opt |SKIP_ZERO ; +break ; +default : +usage (); +exit (1 ); + } + } + +read_alloc_info (); + +} --2.25.1
Hao Ge
reviewerHi Suren and Alessio
Thanks for your information.
Quoted 9 lines · Alessio Balsini
Thank you Suren for looping me in and Hao for the tool!
As Suren mentioned, we recently developed
alloctopfor the same reason as
yours: easy and fast manipulation of/proc/allocinfoinstead of having long chains of piped sorts, greps, awks...alloctoptakesslabtopas a reference for command line parameters and output format.
It currently supports sorting and filtering by minimum allocation size, as well as some kind of aggregation based on tag (similar to whatdudoes).
Yes, I originally intended to gradually add such functionalities as well as those
discovered during subsequent usage.
Quoted 2 lines
As Suren mentioned, the tool is currently being tested on Android, but we would be happy to collaborate on adding new features and making it part of
tools/mm.
I'm also happy to do this.
I salute everyone who has put in the effort for this.
Make this tool increasingly user-friendly.
Thanks Best Regards Hao
Quoted 246 lines
Thanks,
AlessioQuoted 241 lines · Suren Baghdasaryan
Quoted 20 lines · Hao Ge
From: Hao Ge<[email protected]>
Some users always say that the information provided by
/proc/allocinfois too extensive or bulky.However, from allocinfo's own perspective, it is not at fault as it needs to provide comprehensive information. Users can then select the information that is useful to them based on their own needs.
For commonly used scenarios, we can develop a tool to facilitate users in filtering and utilizing relevant information from allocinfo.
Currently, there is only one filtering feature available, which is to filter out entries where the 'bytes' field is 0 (since it is often used to understand the current memory allocation status, as previously mentioned by David Wang, who noticed that 2/3 of the lines have an accumulative counter of 0, indicating no memory activities,it will fill up the entire screen,by using the "-s" parameter, a lot of unnecessary information for this scenario can be filtered out.)
Quoted 3 lines
In subsequent phases, we will continue to add more features to this tool, with the goal of making it convenient for most people to use the memory allocation profiling tool.
CC'ing Alessio along with Pasha and Sourav who were interested in such a tool.
Hi Hao,
Thanks for the tool! Actually Alessio just developed a tool called alloctop (similar to slabtop) which I think will do what you want and more. It supports sorting, filtering, continuous update, etc. It's written in Rust and we are planning to upstream it once we finish testing and evaluating it on Android. Please take a look and see if it fits your usecase. Please also note that this tool has been implemented just last week, so hot off the press and might have some early bugs.
Thanks,
Suren.Quoted 201 lines
Signed-off-by Hao Ge<[email protected]>--- MAINTAINERS | 1 + tools/mm/Makefile | 4 +- tools/mm/allocinfo_tool.c | 150 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 153 insertions(+), 2 deletions(-)create mode 100644
tools/mm/allocinfo_tool.cMAINTAINERS +1
-0@@ -15000,6 +15000,7 @@ F: Documentation/mm/allocation-profiling.rst F: include/linux/alloc_tag.h F: include/linux/pgalloc_tag.h F: lib/alloc_tag.c +F: tools/mm/allocinfo_tool.c MEMORY CONTROLLER DRIVERS M: Krzysztof Kozlowski<[email protected]>tools/mm/Makefile +2
-2@@ -3,7 +3,7 @@ # include ../scripts/Makefile.include -BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test +BUILD_TARGETS=page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool INSTALL_TARGETS = $(BUILD_TARGETS) thpmaps LIB_DIR = ../lib/api@@ -23,7 +23,7 @@ $(LIBS): $(CC) $(CFLAGS) -o $@ $< $(LDFLAGS) clean: - $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test + $(RM) page-types slabinfo page_owner_sort thp_swap_allocator_test allocinfo_tool make -C $(LIB_DIR) clean sbindir ?= /usr/sbintools/mm/allocinfo_tool.c new file +150
-1@@ -0,0 +1,150 @@ +// SPDX-License-Identifier: GPL-2.0 +/* +* allocinfo_tool: Tool to parse allocinfo +* +* Authors: Hao Ge<[email protected]> +* +* Compile with: +* gcc -o allocinfo_tool allocinfo_tool.c +*/ + +#include <stdio.h> +#include <stdlib.h> +#include <sys/stat.h> +#include <string.h> +#include <getopt.h> + +#define ALLOCINFO_FILE "/proc/allocinfo" +#define BUF_SIZE 1024 +#define NAME_MAX_LENTH 64 + +struct alloc_tag_counters { +signed long long bytes; +unsigned long long calls; +}; + +struct codetag { +unsigned int lineno; +char modname[NAME_MAX_LENTH ]; +char function[NAME_MAX_LENTH ]; +char filename[NAME_MAX_LENTH ]; +}; + +struct alloc_info { +struct alloc_tag_counters counters; +struct codetag tag; +int has_modname; +}; + +static int arg_opt; + +enum OPT_BIT { +SKIP_ZERO =1 <<0 , +}; + +void usage (void ) +{ +printf ("Usage: ./allocinfo_tool [OPTIONS]\n" +"-s\t\t\tskip bytes for 0 allocinfo entries\n" + ); +} + +int parse_alloc_info (char *line,struct alloc_info *info) +{ +if (sscanf (line,"%12lli %8llu %[^:]:%d [%[^]]] func:%s" , + &info->counters.bytes, &info->counters.calls, + info->tag.filename, &info->tag.lineno, + info->tag.modname, info->tag.function) ==6 ){ + info->has_modname =1 ; +return 1 ; + }; + +if (sscanf (line,"%12llu %8llu %[^:]:%u func:%s" , + &info->counters.bytes, &info->counters.calls, + info->tag.filename, &info->tag.lineno, + info->tag.function) ==5 ){ + info->has_modname =0 ; +return 1 ; + } + +return 0 ; +} + +int read_alloc_info (void ) +{ +FILE *file =fopen (ALLOCINFO_FILE ,"r" ); + +if (!file) { +perror ("Failed to open /proc/allocinfo" ); +return EXIT_FAILURE ; + } + +int line =0 , i =0 ; +char *buffer =malloc (BUF_SIZE ); +struct alloc_info *info; + +while (fgets (buffer,BUF_SIZE , file)) { + +/* +* allocinfo - version: 1.0 +* # <size> <calls> <tag info> +*/ +if (line <2 ) { +printf ("%s" , buffer); + line++; +continue ; + } + + info =realloc (info,sizeof (struct alloc_info ) * (i +1 )); + +if (parse_alloc_info (buffer, info + i) ==0 ) { +printf ("Mismatch with the format of /proc/allocinfo" ); +return 0 ; + } + +if ((arg_opt &SKIP_ZERO ) && (info[i].counters.bytes ==0 )) +continue ; + +printf ("%12lli %8llu " , info[i].counters.bytes, info[i].counters.calls); + +if (info[i].has_modname) +printf ("%s:%u [%s] func:%s" , + info[i].tag.filename, info[i].tag.lineno, + info[i].tag.modname, info[i].tag.function); +else +printf ("%s:%u func:%s" , + info[i].tag.filename, info[i].tag.lineno, + info[i].tag.function); +printf (" " ); +printf ("\n" ); + i++; + } + +free (info); +free (buffer); +fclose (file); +} + +int main (int argc,char *argv[]) +{ + +int opt; +struct option longopts[] = { + {"s" ,0 ,NULL ,1 }, + {0 ,0 ,0 ,0 }, + }; + +while ((opt =getopt_long (argc, argv,"s" , longopts,NULL )) !=-1 ) { +switch (opt) { +case 's' : + arg_opt = arg_opt |SKIP_ZERO ; +break ; +default : +usage (); +exit (1 ); + } + } + +read_alloc_info (); + +} --2.25.1
Hi,
I have using this feature for a long while, and I believe this memory alloc profiling feature is quite powerful.
But, I have been wondering how to use this data, specifically:
how anomaly could be detected, what pattern should be defined as anomaly?
So far, I have tools collecting those data (via prometheus), make basic analysis, i.e. top-k, group-by or rate. Those analysis help me understand my system, but I cannot tell whether it is abnormal or not.
And sometimes I would just read through /proc/allocinfo, trying to pickup something. (Sometimes get lucky, actually only once, find the underflow problem weeks ago.)
A tool would be more helpful if it can identify anomalies, and we can add more pattern as develop along.
A pattern may be hard to define, especially when it involves context. For example, I happened to notice following strange things recently:
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 1025
896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 1025
96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 1025
12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 1025 <----- B
0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 210
0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 210
0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 210
0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 210 <----- ACode A
2230 sdd->sd = alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return -ENOMEM;
2233
Code B
2246 for_each_cpu(j, cpu_map) {
...2251
2252 sd = kzalloc_node(sizeof(struct sched_domain) + cpumask_size(),
2253 GFP_KERNEL, cpu_to_node(j));
2254 if (!sd)
2255 return -ENOMEM;
2256
2257 *per_cpu_ptr(sdd->sd, j) = sd;
The address of memory alloced by 'Code B', is stored in memory "Code A', the allocation counter for 'Code A' is *0*, while 'Code B' is not *0*. Something odd happens here, either it is expected and some ownership changes happened somewhere , or it is a leak, or it is an accounting problem.
If a tool can help identify this kind of pattern, that would be great!~
Any suggestions about how to proceed with the memory problem of kernel/sched/topology.c mentioneded
above?, or is it a problem at all?
Thanks
David
Quoted 25 lines · Suren Baghdasaryan
Quoted 6 lines · Hao Ge
From: Hao Ge <[email protected]>
Some users always say that the information provided by
/proc/allocinfois too extensive or bulky.CC'ing Alessio along with Pasha and Sourav who were interested in such a tool.
Hi Hao,
Thanks for the tool! Actually Alessio just developed a tool called alloctop (similar to slabtop) which I think will do what you want and more. It supports sorting, filtering, continuous update, etc. It's written in Rust and we are planning to upstream it once we finish testing and evaluating it on Android. Please take a look and see if it fits your usecase. Please also note that this tool has been implemented just last week, so hot off the press and might have some early bugs.
Thanks,
Suren.Quoted 1 line
Quoted 50 lines · David Wang
Hi,
I have using this feature for a long while, and I believe this memory alloc profiling feature is quite powerful.
But, I have been wondering how to use this data, specifically:
how anomaly could be detected, what pattern should be defined as anomaly?So far, I have tools collecting those data (via prometheus), make basic analysis, i.e. top-k, group-by or rate. Those analysis help me understand my system, but I cannot tell whether it is abnormal or not.
And sometimes I would just read through
/proc/allocinfo, trying to pickup something. (Sometimes get lucky, actually only once, find the underflow problem weeks ago.)A tool would be more helpful if it can identify anomalies, and we can add more pattern as develop along.
A pattern may be hard to define, especially when it involves context. For example, I happened to notice following strange things recently:
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 1025 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 1025 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 1025 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 1025 <----- B 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 210 <----- ACode A
2230sdd->sd=alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return-ENOMEM;
2233Code B
2246 for_each_cpu(j, cpu_map) { ...2251
2252 sd = kzalloc_node(sizeof(structsched_domain) +cpumask_size(),
2253GFP_KERNEL,cpu_to_node(j));
2254 if (!sd)
2255 return-ENOMEM;
2256
2257 *per_cpu_ptr(sdd->sd, j) = sd;The address of memory alloced by 'Code B', is stored in memory "Code A', the allocation counter for 'Code A' is *0*, while 'Code B' is not *0*. Something odd happens here, either it is expected and some ownership changes happened somewhere , or it is a leak, or it is an accounting problem.
Quoted 6 lines
If a tool can help identify this kind of pattern, that would be great!~
Any suggestions about how to proceed with the memory problem of
kernel/sched/topology.cmentioneded above?, or is it a problem at all?
Update:
It seems the memory alloced by 'Code B' could be handovered via claim_allocations:
1530 /*
1531 * NULL the sd_data elements we've used to build the sched_domain and 1532 * sched_group structure so that the subsequent __free_domain_allocs() 1533 * will not free the data we're using. 1534 */
1535 static void claim_allocations(int cpu, struct sched_domain *sd)
So most likely, this is neither a leak nor a accounting issue. False alarm, sorry....
The reason I brought this up is that the profiling data is rich, but a user who is not familiar with code detail could not make much out of it. If a tool can tell whether the system is drifting away somewhere, like a healthcheck based on profiling data, it would be quite helpful.
Thanks
David
memory alloc profiling seems not work properly during bootup?
Hi,
More update,
When I boot up my system, no alloc_percpu was accounted in kernel/sched/topology.c
996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80
996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80
96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80
12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80
612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1And then after suspend/resume, those alloc_percpu shows up.
996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395
996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395
96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395
12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395
0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70 <---
0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70 <---
0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70 <---
0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70 <---
612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1I have my accumulative counter patch and filter out items with 0 accumulative counter, I am almost sure the patch would not cause this accounting issue, but not 100%.....
It seems to me, during boot up, some alloc_percpu is not registered.
FYI
David
Quoted 80 lines · David Wang
Quoted 50 lines · David Wang
Hi,
I have using this feature for a long while, and I believe this memory alloc profiling feature is quite powerful.
But, I have been wondering how to use this data, specifically:
how anomaly could be detected, what pattern should be defined as anomaly?So far, I have tools collecting those data (via prometheus), make basic analysis, i.e. top-k, group-by or rate. Those analysis help me understand my system, but I cannot tell whether it is abnormal or not.
And sometimes I would just read through
/proc/allocinfo, trying to pickup something. (Sometimes get lucky, actually only once, find the underflow problem weeks ago.)A tool would be more helpful if it can identify anomalies, and we can add more pattern as develop along.
A pattern may be hard to define, especially when it involves context. For example, I happened to notice following strange things recently:
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 1025 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 1025 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 1025 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 1025 <----- B 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 210 <----- ACode A
2230sdd->sd=alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return-ENOMEM;
2233Code B
2246 for_each_cpu(j, cpu_map) { ...2251
2252 sd = kzalloc_node(sizeof(structsched_domain) +cpumask_size(),
2253GFP_KERNEL,cpu_to_node(j));
2254 if (!sd)
2255 return-ENOMEM;
2256
2257 *per_cpu_ptr(sdd->sd, j) = sd;The address of memory alloced by 'Code B', is stored in memory "Code A', the allocation counter for 'Code A' is *0*, while 'Code B' is not *0*. Something odd happens here, either it is expected and some ownership changes happened somewhere , or it is a leak, or it is an accounting problem.
Quoted 6 lines
If a tool can help identify this kind of pattern, that would be great!~
Any suggestions about how to proceed with the memory problem of
kernel/sched/topology.cmentioneded above?, or is it a problem at all?Update:
It seems the memory alloced by 'Code B' could be handovered via
claim_allocations:
1530 /*1531 * NULL the sd_data elements we've used to build the sched_domain and 1532 * sched_group structure so that the subsequent __free_domain_allocs() 1533 * will not free the data we're using. 1534 */1535 static void
claim_allocations(int cpu, struct sched_domain *sd)So most likely, this is neither a leak nor a accounting issue. False alarm, sorry....
The reason I brought this up is that the profiling data is rich, but a user who is not familiar with code detail could not make much out of it. If a tool can tell whether the system is drifting away somewhere, like a healthcheck based on profiling data, it would be quite helpful.
Thanks
David
Suren Baghdasaryan
maintainerRe: memory alloc profiling seems not work properly during bootup?
Quoted 25 lines · David Wang
Hi,
More update,
When I boot up my system, no
alloc_percpuwas accounted inkernel/sched/topology.c996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80 996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80 12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80 612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1And then after suspend/resume, those
alloc_percpushows up.996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395 996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395 12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70 <--- 612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1
Quoted 2 lines
I have my accumulative counter patch and filter out items with 0 accumulative counter, I am almost sure the patch would not cause this accounting issue, but not 100%.....
Have you tested this without your accumulative counter patch? IIUC, that patch filters out any allocation which has never been hit. So, if suspend/resume path contains allocations which were never hit before then those allocations would become suddenly visible, like in your case. That's why I'm against filtering allocinfo data in the kernel. Please try this without your patch and see if the data becomes more consistent.
Thanks,
Suren.
Quoted 91 lines
It seems to me, during boot up, some
alloc_percpuis not registered.FYI
DavidQuoted 80 lines · David Wang
Quoted 50 lines · David Wang
Hi,
I have using this feature for a long while, and I believe this memory alloc profiling feature is quite powerful.
But, I have been wondering how to use this data, specifically:
how anomaly could be detected, what pattern should be defined as anomaly?So far, I have tools collecting those data (via prometheus), make basic analysis, i.e. top-k, group-by or rate. Those analysis help me understand my system, but I cannot tell whether it is abnormal or not.
And sometimes I would just read through
/proc/allocinfo, trying to pickup something. (Sometimes get lucky, actually only once, find the underflow problem weeks ago.)A tool would be more helpful if it can identify anomalies, and we can add more pattern as develop along.
A pattern may be hard to define, especially when it involves context. For example, I happened to notice following strange things recently:
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 1025 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 1025 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 1025 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 1025 <----- B 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 210 <----- ACode A
2230sdd->sd=alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return-ENOMEM;
2233Code B
2246 for_each_cpu(j, cpu_map) { ...2251
2252 sd = kzalloc_node(sizeof(structsched_domain) +cpumask_size(),
2253GFP_KERNEL,cpu_to_node(j));
2254 if (!sd)
2255 return-ENOMEM;
2256
2257 *per_cpu_ptr(sdd->sd, j) = sd;The address of memory alloced by 'Code B', is stored in memory "Code A', the allocation counter for 'Code A' is *0*, while 'Code B' is not *0*. Something odd happens here, either it is expected and some ownership changes happened somewhere , or it is a leak, or it is an accounting problem.
Quoted 6 lines
If a tool can help identify this kind of pattern, that would be great!~
Any suggestions about how to proceed with the memory problem of
kernel/sched/topology.cmentioneded above?, or is it a problem at all?Update:
It seems the memory alloced by 'Code B' could be handovered via
claim_allocations:
1530 /*1531 * NULL the sd_data elements we've used to build the sched_domain and 1532 * sched_group structure so that the subsequent __free_domain_allocs() 1533 * will not free the data we're using. 1534 */1535 static void
claim_allocations(int cpu, struct sched_domain *sd)So most likely, this is neither a leak nor a accounting issue. False alarm, sorry....
The reason I brought this up is that the profiling data is rich, but a user who is not familiar with code detail could not make much out of it. If a tool can tell whether the system is drifting away somewhere, like a healthcheck based on profiling data, it would be quite helpful.
Thanks
David
Re: memory alloc profiling seems not work properly during bootup?
Hi,
Quoted 29 lines · Suren Baghdasaryan
Quoted 25 lines · David Wang
Hi,
More update,
When I boot up my system, no
alloc_percpuwas accounted inkernel/sched/topology.c996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80 996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80 12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80 612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1And then after suspend/resume, those
alloc_percpushows up.996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395 996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395 12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70 <--- 612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1Quoted 2 lines
I have my accumulative counter patch and filter out items with 0 accumulative counter, I am almost sure the patch would not cause this accounting issue, but not 100%.....
Quoted 7 lines
Have you tested this without your accumulative counter patch? IIUC, that patch filters out any allocation which has never been hit. So, if suspend/resume path contains allocations which were never hit before then those allocations would become suddenly visible, like in your case. That's why I'm against filtering allocinfo data in the kernel. Please try this without your patch and see if the data becomes more consistent.
I remove all my patch and build a 6.13.0-rc7 kernel, After boot up,
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc
896 14 kernel/sched/topology.c:2266 func:__sdt_alloc
96 6 kernel/sched/topology.c:2259 func:__sdt_alloc
12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc
0 0 kernel/sched/topology.c:2242 func:__sdt_alloc
0 0 kernel/sched/topology.c:2238 func:__sdt_alloc
0 0 kernel/sched/topology.c:2234 func:__sdt_alloc
0 0 kernel/sched/topology.c:2230 func:__sdt_alloc
512 1 kernel/sched/topology.c:1961 func:sched_init_numa
And after suspend/resume, no change detected:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc
896 14 kernel/sched/topology.c:2266 func:__sdt_alloc
96 6 kernel/sched/topology.c:2259 func:__sdt_alloc
12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc
0 0 kernel/sched/topology.c:2242 func:__sdt_alloc
0 0 kernel/sched/topology.c:2238 func:__sdt_alloc
0 0 kernel/sched/topology.c:2234 func:__sdt_alloc
0 0 kernel/sched/topology.c:2230 func:__sdt_alloc
512 1 kernel/sched/topology.c:1961 func:sched_init_numa
I also build a image with accumulative counter, but no filter.
After boot up:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 2
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80
896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80
96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80
12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80
0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 0 <---this *0* seems wrong
0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 0
0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 0
0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 0
512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1 And then suspend/resume:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 17
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395
896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395
96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395
12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395
0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70
0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70
0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70
0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70
512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1 Reading the code, those allocation behaviors should be tied together:
if kzalloc_node at line#2252 happened, then alloc_percpu at line#2230 should also happened.
kernel/sched/topology.c
2230 sdd->sd = alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return -ENOMEM;
...
2246 for_each_cpu(j, cpu_map) {
...
2252 sd = kzalloc_node(sizeof(struct sched_domain) + cpumask_size(),
2253 GFP_KERNEL, cpu_to_node(j));
...
2257 *per_cpu_ptr(sdd->sd, j) = sd;
But somehow during bootup, those alloc_percpu in kernel/sched/topology.c:__sdt_alloc were missed in profiling. (I am not meant to sell the idea of accumulative counter again here, but it dose help sometimes. :).
Quoted 5 lines
Thanks,
Suren.Quoted 1 line
Thanks
David
Suren Baghdasaryan
maintainerRe: memory alloc profiling seems not work properly during bootup?
Quoted 92 lines · David Wang
Hi,
Quoted 29 lines · Suren Baghdasaryan
Quoted 25 lines · David Wang
Hi,
More update,
When I boot up my system, no
alloc_percpuwas accounted inkernel/sched/topology.c996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80 996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80 12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80 612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1And then after suspend/resume, those
alloc_percpushows up.996 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395 996 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395 12388 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70 <--- 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70 <--- 612 1 kernel/sched/topology.c:1961 func:sched_init_numa 1Quoted 2 lines
I have my accumulative counter patch and filter out items with 0 accumulative counter, I am almost sure the patch would not cause this accounting issue, but not 100%.....
Quoted 7 lines
Have you tested this without your accumulative counter patch? IIUC, that patch filters out any allocation which has never been hit. So, if suspend/resume path contains allocations which were never hit before then those allocations would become suddenly visible, like in your case. That's why I'm against filtering allocinfo data in the kernel. Please try this without your patch and see if the data becomes more consistent.
I remove all my patch and build a 6.13.0-rc7 kernel, After boot up,
64 1kernel/sched/topology.c:2579func:alloc_sched_domains
896 14kernel/sched/topology.c:2275func:__sdt_alloc
896 14kernel/sched/topology.c:2266func:__sdt_alloc
96 6kernel/sched/topology.c:2259func:__sdt_alloc
12288 24kernel/sched/topology.c:2252func:__sdt_alloc
0 0kernel/sched/topology.c:2242func:__sdt_alloc
0 0kernel/sched/topology.c:2238func:__sdt_alloc
0 0kernel/sched/topology.c:2234func:__sdt_alloc
0 0kernel/sched/topology.c:2230func:__sdt_alloc
512 1kernel/sched/topology.c:1961func:sched_init_numaAnd after suspend/resume, no change detected:
64 1kernel/sched/topology.c:2579func:alloc_sched_domains
896 14kernel/sched/topology.c:2275func:__sdt_alloc
896 14kernel/sched/topology.c:2266func:__sdt_alloc
96 6kernel/sched/topology.c:2259func:__sdt_alloc
12288 24kernel/sched/topology.c:2252func:__sdt_alloc
0 0kernel/sched/topology.c:2242func:__sdt_alloc
0 0kernel/sched/topology.c:2238func:__sdt_alloc
0 0kernel/sched/topology.c:2234func:__sdt_alloc
0 0kernel/sched/topology.c:2230func:__sdt_alloc
512 1kernel/sched/topology.c:1961func:sched_init_numaI also build a image with accumulative counter, but no filter.
After boot up:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 2 896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 0 <---this *0* seems wrong 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 0 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 0 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 0 512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1And then suspend/resume:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 17 896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70 512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1>
Quoted 2 lines
Reading the code, those allocation behaviors should be tied together:
ifkzalloc_nodeat line#2252 happened, thenalloc_percpuat line#2230 should also happened.
Hmm, ok. Looks like early calls to alloc_percpu() are not being registered somehow. Could you please share your cumulative counter patch with me? I'll try to reproduce this locally and see if I can spot the issue.
Quoted 25 lines
kernel/sched/topology.c
2230sdd->sd=alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return-ENOMEM;
...
2246for_each_cpu(j, cpu_map){
...
2252 sd = kzalloc_node(sizeof(structsched_domain) +cpumask_size(),
2253GFP_KERNEL,cpu_to_node(j));
...
2257 *per_cpu_ptr(sdd->sd, j) = sd;But somehow during bootup, those
alloc_percpuin kernel/sched/topology.c:__sdt_alloc were missed in profiling. (I am not meant to sell the idea of accumulative counter again here, but it dose help sometimes. :).Quoted 5 lines
Thanks,
Suren.Quoted 1 line
Thanks
David
Re: memory alloc profiling seems not work properly during bootup?
Quoted 67 lines · Suren Baghdasaryan
Quoted 63 lines · David Wang
Quoted 5 lines
Quoted 4 lines
I have my accumulative counter patch and filter out items with 0 accumulative counter, I am almost sure the patch would not cause this accounting issue, but not 100%.....
Quoted 7 lines
Have you tested this without your accumulative counter patch? IIUC, that patch filters out any allocation which has never been hit. So, if suspend/resume path contains allocations which were never hit before then those allocations would become suddenly visible, like in your case. That's why I'm against filtering allocinfo data in the kernel. Please try this without your patch and see if the data becomes more consistent.
I remove all my patch and build a 6.13.0-rc7 kernel, After boot up,
64 1kernel/sched/topology.c:2579func:alloc_sched_domains
896 14kernel/sched/topology.c:2275func:__sdt_alloc
896 14kernel/sched/topology.c:2266func:__sdt_alloc
96 6kernel/sched/topology.c:2259func:__sdt_alloc
12288 24kernel/sched/topology.c:2252func:__sdt_alloc
0 0kernel/sched/topology.c:2242func:__sdt_alloc
0 0kernel/sched/topology.c:2238func:__sdt_alloc
0 0kernel/sched/topology.c:2234func:__sdt_alloc
0 0kernel/sched/topology.c:2230func:__sdt_alloc
512 1kernel/sched/topology.c:1961func:sched_init_numaAnd after suspend/resume, no change detected:
64 1kernel/sched/topology.c:2579func:alloc_sched_domains
896 14kernel/sched/topology.c:2275func:__sdt_alloc
896 14kernel/sched/topology.c:2266func:__sdt_alloc
96 6kernel/sched/topology.c:2259func:__sdt_alloc
12288 24kernel/sched/topology.c:2252func:__sdt_alloc
0 0kernel/sched/topology.c:2242func:__sdt_alloc
0 0kernel/sched/topology.c:2238func:__sdt_alloc
0 0kernel/sched/topology.c:2234func:__sdt_alloc
0 0kernel/sched/topology.c:2230func:__sdt_alloc
512 1kernel/sched/topology.c:1961func:sched_init_numaI also build a image with accumulative counter, but no filter.
After boot up:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 2 896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 0 <---this *0* seems wrong 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 0 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 0 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 0 512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1And then suspend/resume:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 17 896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70 512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1>Quoted 2 lines
Reading the code, those allocation behaviors should be tied together:
ifkzalloc_nodeat line#2252 happened, thenalloc_percpuat line#2230 should also happened.
Quoted 5 lines
Hmm, ok. Looks like early calls to
alloc_percpu()are not being registered somehow. Could you please share your cumulative counter patch with me? I'll try to reproduce this locally and see if I canspot the issue.
Sure, here is the patch base on 6.13.0-rc7.
include/linux/alloc_tag.h +0 -0
@@ -18,6 +18,7 @@
struct alloc_tag_counters {
u64 bytes;
u64 calls;
+ u64 accu_calls;
};
/*
@@ -124,7 +125,7 @@ static inline bool mem_alloc_profiling_enabled(void)
static inline struct alloc_tag_counters alloc_tag_read(struct alloc_tag *tag)
{
- struct alloc_tag_counters v = { 0, 0 };
+ struct alloc_tag_counters v = { 0, 0, 0 };
struct alloc_tag_counters *counter;
int cpu;
@@ -132,6 +133,7 @@ static inline struct alloc_tag_counters alloc_tag_read(struct alloc_tag *tag) counter = per_cpu_ptr(tag->counters, cpu);
v.bytes += counter->bytes;
v.calls += counter->calls;
+ v.accu_calls += counter->accu_calls;
}
return v;
@@ -179,6 +181,7 @@ static inline bool alloc_tag_ref_set(union codetag_ref *ref, struct alloc_tag *t
* counter because when we free each part the counter will be decremented.
*/
this_cpu_inc(tag->counters->calls);
+ this_cpu_inc(tag->counters->accu_calls);
return true;
}
lib/alloc_tag.c +0 -0
@@ -97,6 +97,7 @@ static void alloc_tag_to_text(struct seq_buf *out, struct codetag *ct)
seq_buf_printf(out, "%12lli %8llu ", bytes, counter.calls); codetag_to_text(out, ct);
+ seq_buf_printf(out, " %llu", counter.accu_calls); seq_buf_putc(out, ' ');
seq_buf_putc(out, '\n');
}
David
Quoted 26 lines
Quoted 26 lines
kernel/sched/topology.c
2230sdd->sd=alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return-ENOMEM;
...
2246for_each_cpu(j, cpu_map){
...
2252 sd = kzalloc_node(sizeof(structsched_domain) +cpumask_size(),
2253GFP_KERNEL,cpu_to_node(j));
...
2257 *per_cpu_ptr(sdd->sd, j) = sd;But somehow during bootup, those
alloc_percpuin kernel/sched/topology.c:__sdt_alloc were missed in profiling. (I am not meant to sell the idea of accumulative counter again here, but it dose help sometimes. :).Quoted 5 lines
Thanks,
Suren.Quoted 1 line
Thanks
David
Suren Baghdasaryan
maintainerRe: memory alloc profiling seems not work properly during bootup?
Quoted 70 lines · David Wang
Quoted 67 lines · Suren Baghdasaryan
Quoted 65 lines · David Wang
Quoted 5 lines
Quoted 4 lines
I have my accumulative counter patch and filter out items with 0 accumulative counter, I am almost sure the patch would not cause this accounting issue, but not 100%.....
Quoted 7 lines
Have you tested this without your accumulative counter patch? IIUC, that patch filters out any allocation which has never been hit. So, if suspend/resume path contains allocations which were never hit before then those allocations would become suddenly visible, like in your case. That's why I'm against filtering allocinfo data in the kernel. Please try this without your patch and see if the data becomes more consistent.
I remove all my patch and build a 6.13.0-rc7 kernel, After boot up,
64 1kernel/sched/topology.c:2579func:alloc_sched_domains
896 14kernel/sched/topology.c:2275func:__sdt_alloc
896 14kernel/sched/topology.c:2266func:__sdt_alloc
96 6kernel/sched/topology.c:2259func:__sdt_alloc
12288 24kernel/sched/topology.c:2252func:__sdt_alloc
0 0kernel/sched/topology.c:2242func:__sdt_alloc
0 0kernel/sched/topology.c:2238func:__sdt_alloc
0 0kernel/sched/topology.c:2234func:__sdt_alloc
0 0kernel/sched/topology.c:2230func:__sdt_alloc
512 1kernel/sched/topology.c:1961func:sched_init_numaAnd after suspend/resume, no change detected:
64 1kernel/sched/topology.c:2579func:alloc_sched_domains
896 14kernel/sched/topology.c:2275func:__sdt_alloc
896 14kernel/sched/topology.c:2266func:__sdt_alloc
96 6kernel/sched/topology.c:2259func:__sdt_alloc
12288 24kernel/sched/topology.c:2252func:__sdt_alloc
0 0kernel/sched/topology.c:2242func:__sdt_alloc
0 0kernel/sched/topology.c:2238func:__sdt_alloc
0 0kernel/sched/topology.c:2234func:__sdt_alloc
0 0kernel/sched/topology.c:2230func:__sdt_alloc
512 1kernel/sched/topology.c:1961func:sched_init_numaI also build a image with accumulative counter, but no filter.
After boot up:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 2 896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 80 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 80 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 80 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 80 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 0 <---this *0* seems wrong 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 0 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 0 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 0 512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1And then suspend/resume:
64 1 kernel/sched/topology.c:2579 func:alloc_sched_domains 17 896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 395 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 395 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 395 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 395 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 70 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 70 512 1 kernel/sched/topology.c:1961 func:sched_init_numa 1>Reading the code, those allocation behaviors should be tied together:
ifkzalloc_nodeat line#2252 happened, thenalloc_percpuat line#2230 should also happened.
Quoted 7 lines
Quoted 5 lines
Hmm, ok. Looks like early calls to
alloc_percpu()are not being registered somehow. Could you please share your cumulative counter patch with me? I'll try to reproduce this locally and see if I canspot the issue.
Sure, here is the patch base on 6.13.0-rc7.
Thanks and sorry for the delay.
It looks like the per-cpu allocations you pointed out happen early enough in the boot process that chunk->obj_exts was not allocated yet. Therefore the check inside pcpu_alloc_tag_alloc_hook() for chunk->obj_exts fails and accounting gets skipped. Allocating obj_exts earlier is not trivial because slab is not available yet. I'll need to look closer into per-cpu code to see how this can be fixed.
Quoted 84 lines
include/linux/alloc_tag.h +4
-1@@ -18,6 +18,7 @@struct alloc_tag_counters {u64 bytes;u64 calls; +u64 accu_calls; };/* @@ -124,7 +125,7 @@ static inline bool mem_alloc_profiling_enabled(void)static inline struct alloc_tag_counters alloc_tag_read (struct alloc_tag *tag) { -struct alloc_tag_counters v = {0 ,0 }; +struct alloc_tag_counters v = {0 ,0 ,0 };struct alloc_tag_counters *counter;int cpu;@@ -132,6 +133,7 @@ static inline struct alloc_tag_counters alloc_tag_read(struct alloc_tag *tag) counter =per_cpu_ptr (tag->counters, cpu); v.bytes += counter->bytes; v.calls += counter->calls; + v.accu_calls += counter->accu_calls; }return v;@@ -179,6 +181,7 @@ static inline bool alloc_tag_ref_set(union codetag_ref *ref, struct alloc_tag *t* counter because when we free each part the counter will be decremented. */ this_cpu_inc (tag->counters->calls); +this_cpu_inc (tag->counters->accu_calls);return true; }lib/alloc_tag.c +1
-0@@ -97,6 +97,7 @@ static void alloc_tag_to_text(struct seq_buf *out, struct codetag *ct)seq_buf_printf (out,"%12lli %8llu " , bytes, counter.calls);codetag_to_text (out, ct); +seq_buf_printf (out," %llu" , counter.accu_calls);seq_buf_putc (out,' ' );seq_buf_putc (out,'\n' ); }David
Quoted 26 lines
Quoted 26 lines
kernel/sched/topology.c
2230sdd->sd=alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return-ENOMEM;
...
2246for_each_cpu(j, cpu_map){
...
2252 sd = kzalloc_node(sizeof(structsched_domain) +cpumask_size(),
2253GFP_KERNEL,cpu_to_node(j));
...
2257 *per_cpu_ptr(sdd->sd, j) = sd;But somehow during bootup, those
alloc_percpuin kernel/sched/topology.c:__sdt_alloc were missed in profiling. (I am not meant to sell the idea of accumulative counter again here, but it dose help sometimes. :).Quoted 5 lines
Thanks,
Suren.Quoted 1 line
Thanks
David
Suren Baghdasaryan
maintainerQuoted 2 lines · David Wang
Hi,
Hi David,
Sorry for the delay. I'm not ignoring your input, I'm just a bit busy and didn't have time to properly reply to your questions.
Quoted 7 lines
I have using this feature for a long while, and I believe this memory alloc profiling feature is quite powerful.
But, I have been wondering how to use this data, specifically:
how anomaly could be detected, what pattern should be defined as anomaly?
Quoted 7 lines
So far, I have tools collecting those data (via prometheus), make basic analysis, i.e. top-k, group-by or rate. Those analysis help me understand my system, but I cannot tell whether it is abnormal or not.
And sometimes I would just read through
/proc/allocinfo, trying to pickup something. (Sometimes get lucky, actually only once, find the underflow problem weeks ago.)A tool would be more helpful if it can identify anomalies, and we can add more pattern as develop along.
You are absolutely correct. An automatic detection of problematic patterns would be the ultimate goal. We are analyzing the data we collect and trying to come up with strategies for identifying such patterns. Simple and obvious pattern for a leak would be constant growth but there might be others like sawtooth pattern or spikes which could point to opportunities to optimize the usage by employing object pools/caches. Categorizing allocations into hierarchical groups and measuring per-group consumption might be another useful technique we are considering. All this is in quite early stages, so ideas and suggestions from people using this API would be very valuable.
Quoted 31 lines
A pattern may be hard to define, especially when it involves context. For example, I happened to notice following strange things recently:
896 14 kernel/sched/topology.c:2275 func:__sdt_alloc 1025 896 14 kernel/sched/topology.c:2266 func:__sdt_alloc 1025 96 6 kernel/sched/topology.c:2259 func:__sdt_alloc 1025 12288 24 kernel/sched/topology.c:2252 func:__sdt_alloc 1025 <----- B 0 0 kernel/sched/topology.c:2242 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2238 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2234 func:__sdt_alloc 210 0 0 kernel/sched/topology.c:2230 func:__sdt_alloc 210 <----- ACode A
2230sdd->sd=alloc_percpu(struct sched_domain *);
2231 if (!sdd->sd)
2232 return-ENOMEM;
2233Code B
2246 for_each_cpu(j, cpu_map) { ...2251
2252 sd = kzalloc_node(sizeof(structsched_domain) +cpumask_size(),
2253GFP_KERNEL,cpu_to_node(j));
2254 if (!sd)
2255 return-ENOMEM;
2256
2257 *per_cpu_ptr(sdd->sd, j) = sd;
Quoted 5 lines
The address of memory alloced by 'Code B', is stored in memory "Code A', the allocation counter for 'Code A' is *0*, while 'Code B' is not *0*. Something odd happens here, either it is expected and some ownership changes happened somewhere , or it is a leak, or it is an accounting problem.
If a tool can help identify this kind of pattern, that would be great!~
Hmm. I don't see an easy way to identify such code dependencies from allocinfo data alone. I think that would involve some sophisticated code analysis tooling.
Quoted 4 lines
Any suggestions about how to proceed with the memory problem of
kernel/sched/topology.cmentioneded
above?, or is it a problem at all?
From your follow-up email, it looks like you already found the answer :) Thanks,
Suren.
Quoted 35 lines
Thanks
DavidQuoted 25 lines · Suren Baghdasaryan
Quoted 6 lines · Hao Ge
From: Hao Ge <[email protected]>
Some users always say that the information provided by
/proc/allocinfois too extensive or bulky.CC'ing Alessio along with Pasha and Sourav who were interested in such a tool.
Hi Hao,
Thanks for the tool! Actually Alessio just developed a tool called alloctop (similar to slabtop) which I think will do what you want and more. It supports sorting, filtering, continuous update, etc. It's written in Rust and we are planning to upstream it once we finish testing and evaluating it on Android. Please take a look and see if it fits your usecase. Please also note that this tool has been implemented just last week, so hot off the press and might have some early bugs.
Thanks,
Suren.Quoted 1 line
kernel test robot
build botkernel test robot noticed the following build warnings:
Hi Hao,
kernel test robot noticed the following build warnings:
[auto build test WARNING on akpm-mm/mm-everything]
[also build test WARNING on linus/master v6.13-rc6 next-20250108] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Hao-Ge/tools-mm-Introduce-a-tool-to-handle-entries-in-allocinfo/20250106-192419 base: https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything patch link: https://lore.kernel.org/r/20250106112103.25401-1-hao.ge%40linux.dev patch subject: [PATCH] tools/mm: Introduce a tool to handle entries in allocinfo compiler: clang version 19.1.3 (https://github.com/llvm/llvm-project ab51eccf88f5321e7c60591c5546b254b6afab99) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250109/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All warnings (new ones prefixed by >>):
>> allocinfo_tool.c:126:1: warning: non-void function does not return a value in all control paths [-Wreturn-type]
126 | }
| ^
1 warning generated.
page-types.c:1003:28: warning: operator '?:' has lower precedence than '+'; '+' will be evaluated first [-Wparentheses]
1003 | end = off + sigbus_addr ? sigbus_addr - ptr : 0;
| ~~~~~~~~~~~~~~~~~ ^
page-types.c:1003:28: note: place parentheses around the '+' expression to silence this warning
1003 | end = off + sigbus_addr ? sigbus_addr - ptr : 0;
| ^
| ( )
page-types.c:1003:28: note: place parentheses around the '?:' expression to evaluate it first
1003 | end = off + sigbus_addr ? sigbus_addr - ptr : 0;
| ^
| ( )1 warning generated.
Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki