* [Tarantool-patches] [PATCH luajit v3] memprof: introduce cli flag to run dump parser
@ 2022-12-05 18:01 Maxim Kokryashkin via Tarantool-patches
2023-02-07 7:23 ` Sergey Kaplun via Tarantool-patches
0 siblings, 1 reply; 2+ messages in thread
From: Maxim Kokryashkin via Tarantool-patches @ 2022-12-05 18:01 UTC (permalink / raw)
To: tarantool-patches, sergos, skaplun
It is really inconvenient to use a standalone shell script to parse
memprof dump. That is why this commit introduces a CLI flag for tools
to the LuaJIT, so now it is possible to parse memprof dump
as simple as:
```
luajit -tm memprof.bin
```
Closes tarantool/tarantool#5688
---
Changes in v3:
- Fixed comments as per review by Sergey
Branch: https://github.com/tarantool/luajit/tree/fckxorg/gh-5688-cli-for-memprof-parse
PR: https://github.com/tarantool/tarantool/pull/8002
Issue: https://github.com/tarantool/tarantool/issues/5688
CMakeLists.txt | 9 +--
Makefile.original | 7 +-
src/CMakeLists.txt | 5 ++
src/lj_tools_conf.h.in | 7 ++
src/luajit.c | 63 ++++++++++++++++--
.../gh-5688-memprof-cli-flag.test.lua | 64 +++++++++++++++++++
tools/CMakeLists.txt | 2 +
7 files changed, 145 insertions(+), 12 deletions(-)
create mode 100644 src/lj_tools_conf.h.in
create mode 100644 test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua
diff --git a/CMakeLists.txt b/CMakeLists.txt
index c870cce2..97d0d42f 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -263,6 +263,11 @@ endif()
# related compiler and linker flags passed. This should be done
# the right way later.
+# --- Tools --------------------------------------------------------------------
+
+add_subdirectory(tools)
+set(LUAJIT_TOOLS_DIR "${LUAJIT_TOOLS_DIR}")
+
# --- Main source tree ---------------------------------------------------------
add_subdirectory(src)
@@ -271,10 +276,6 @@ add_subdirectory(src)
add_subdirectory(etc)
-# --- Tools --------------------------------------------------------------------
-
-add_subdirectory(tools)
-
# --- Testing source tree ------------------------------------------------------
# Auxiliary options for testing.
diff --git a/Makefile.original b/Makefile.original
index 0c92df9e..bb0ab73d 100644
--- a/Makefile.original
+++ b/Makefile.original
@@ -104,6 +104,7 @@ FILES_UTILSLIB= avl.lua bufread.lua symtab.lua
FILES_MEMPROFLIB= parse.lua humanize.lua
FILES_TOOLSLIB= memprof.lua
FILE_TMEMPROF= luajit-parse-memprof
+FILE_TOOLSHEADER= lj_tools_conf.h.in
ifeq (,$(findstring Windows,$(OS)))
HOST_SYS:= $(shell uname -s)
@@ -193,7 +194,7 @@ clean:
$(RM) tools/$(FILE_TMEMPROF)
$(MAKE) -C src -f Makefile.original clean
-tools: tools/$(FILE_TMEMPROF)
+tools: tools/$(FILE_TMEMPROF) src/$(FILE_TOOLSHEADER)
# FIXME: This is an ugly hack to manually configure an auxiliary
# tools/luajit-parse-memprof. This file should go away in scope of
@@ -204,6 +205,10 @@ tools/$(FILE_TMEMPROF): src/luajit
$@.in > $@
@chmod +x $@
+src/$(FILE_TOOLSHEADER):
+ @sed -e "s|@LUAJIT_TOOLS_DIR@|$(realpath tools)|" \
+ $@.in > $@
+
.PHONY: all install amalg clean tools
##############################################################################
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index dffc0a4d..17674a41 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -145,6 +145,8 @@ make_source_list(SOURCES_CORE_NO_JIT_FFI
${SOURCES_UTILS}
)
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/lj_tools_conf.h.in ${CMAKE_CURRENT_SOURCE_DIR}/lj_tools_conf.h)
+
set(SOURCES_CORE ${SOURCES_CORE_NO_JIT_FFI})
# Build JIT sources if JIT support is enabled.
@@ -251,6 +253,9 @@ add_custom_target(
jit/vmdef.lua
)
+# --- Generate luajit tools config header -------------------------------------
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/lj_tools_conf.h.in ${CMAKE_CURRENT_SOURCE_DIR}/lj_tools_conf.h)
+
# --- Generate core and VM object files ---------------------------------------
# Virtual machine.
diff --git a/src/lj_tools_conf.h.in b/src/lj_tools_conf.h.in
new file mode 100644
index 00000000..9f9a2e49
--- /dev/null
+++ b/src/lj_tools_conf.h.in
@@ -0,0 +1,7 @@
+#ifndef LJ_TOOLS_CONF_H
+#define LJ_TOOLS_CONF_H
+
+#define TOOLS_PATH "@LUAJIT_TOOLS_DIR@/?.lua"
+#define PARSER_PATH "@LUAJIT_TOOLS_DIR@/memprof.lua"
+
+#endif
diff --git a/src/luajit.c b/src/luajit.c
index 1ca24301..bd9ae8f3 100644
--- a/src/luajit.c
+++ b/src/luajit.c
@@ -19,6 +19,8 @@
#include "lj_arch.h"
+#include "lj_tools_conf.h"
+
#if LJ_TARGET_POSIX
#include <unistd.h>
#define lua_stdin_is_tty() isatty(0)
@@ -72,6 +74,7 @@ static void print_usage(void)
" -O[opt] Control LuaJIT optimizations.\n"
" -i Enter interactive mode after executing " LUA_QL("script") ".\n"
" -v Show version information.\n"
+ " -t[cmd] Execute tool.\n"
" -E Ignore environment variables.\n"
" -- Stop handling options.\n"
" - Execute stdin and stop handling options.\n", stderr);
@@ -266,13 +269,9 @@ static void dotty(lua_State *L)
progname = oldprogname;
}
-static int handle_script(lua_State *L, char **argx)
+static int call_script(lua_State *L, const char *fname)
{
- int status;
- const char *fname = argx[0];
- if (strcmp(fname, "-") == 0 && strcmp(argx[-1], "--") != 0)
- fname = NULL; /* stdin */
- status = luaL_loadfile(L, fname);
+ int status = luaL_loadfile(L, fname);
if (status == LUA_OK) {
/* Fetch args from arg table. LUA_INIT or -e might have changed them. */
int narg = 0;
@@ -290,6 +289,16 @@ static int handle_script(lua_State *L, char **argx)
}
status = docall(L, narg, 0);
}
+ return status;
+}
+
+static int handle_script(lua_State *L, char **argx)
+{
+ int status;
+ const char *fname = argx[0];
+ if (strcmp(fname, "-") == 0 && strcmp(argx[-1], "--") != 0)
+ fname = NULL; /* stdin */
+ status = call_script(L, fname);
return report(L, status);
}
@@ -361,6 +370,15 @@ static int dojitcmd(lua_State *L, const char *cmd)
return runcmdopt(L, opt ? opt+1 : opt);
}
+static int dotoolcmd(lua_State *L, const char *cmd)
+{
+ if(strcmp(cmd, "m") == 0) {
+ const int status = call_script(L, PARSER_PATH);
+ return report(L, status);
+ }
+ return -1;
+}
+
/* Optimization flags. */
static int dojitopt(lua_State *L, const char *opt)
{
@@ -390,6 +408,26 @@ static int dobytecode(lua_State *L, char **argv)
return -1;
}
+/*
+** On most Linux distros, it is the default value for the
+** maximum length of a string passed to `execve`.
+** However, there is no common value for other OSes, so
+** the size of 32 default memory pages is adopted.
+**/
+#define MAX_ENV_VAR 32 * 4096
+
+static int update_env_var(const char *name, const char *value)
+{
+ char env_buf[MAX_ENV_VAR] = "";
+ const char *env = getenv(name);
+ if (env == NULL) {
+ return setenv(name, value, 0);
+ } else {
+ strcpy(env_buf, env);
+ return setenv(name, strcat(env_buf, value), 0);
+ }
+}
+
/* check that argument has no extra characters at the end */
#define notail(x) {if ((x)[2] != '\0') return -1;}
@@ -398,6 +436,7 @@ static int dobytecode(lua_State *L, char **argv)
#define FLAGS_EXEC 4
#define FLAGS_OPTION 8
#define FLAGS_NOENV 16
+#define FLAGS_TOOL 32
static int collectargs(char **argv, int *flags)
{
@@ -419,6 +458,12 @@ static int collectargs(char **argv, int *flags)
notail(argv[i]);
*flags |= FLAGS_VERSION;
break;
+ case 't':
+ *flags |= FLAGS_TOOL;
+ if (argv[i][2] == '\0') return -1;
+ if (argv[i + 1] == NULL) return -1;
+ update_env_var("LUA_PATH", TOOLS_PATH);
+ return i + 1;
case 'e':
*flags |= FLAGS_EXEC;
case 'j': /* LuaJIT extension */
@@ -474,6 +519,10 @@ static int runargs(lua_State *L, char **argv, int argn)
return 1;
break;
}
+ case 't': { /* Tarantool's fork extension. */
+ const char *cmd = argv[i] + 2;
+ return dotoolcmd(L, cmd) == LUA_OK;
+ }
case 'O': /* LuaJIT extension. */
if (dojitopt(L, argv[i] + 2))
return 1;
@@ -535,7 +584,7 @@ static int pmain(lua_State *L)
luaL_openlibs(L);
lua_gc(L, LUA_GCRESTART, -1);
- createargtable(L, argv, s->argc, argn);
+ createargtable(L, argv, s->argc, (flags & FLAGS_TOOL) ? argn - 1 : argn);
if (!(flags & FLAGS_NOENV)) {
s->status = handle_luainit(L);
diff --git a/test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua b/test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua
new file mode 100644
index 00000000..ba2d0219
--- /dev/null
+++ b/test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua
@@ -0,0 +1,64 @@
+local utils = require('utils')
+
+-- XXX: The patch is for luajit only, and it doesn't
+-- work on Tarantool.
+-- luacheck: no global
+utils.skipcond(
+ (jit.arch ~= 'x86' and jit.arch ~= 'x64') or _TARANTOOL,
+ jit.arch..' architecture is NIY for memprof'
+)
+
+local tap = require('tap')
+
+local test = tap.test('gh-5688-memprof-cli-flag')
+test:plan(2)
+
+jit.off()
+jit.flush()
+
+local table_new = require 'table.new'
+
+local TMP_BINFILE = utils.profilename('memprofdata.tmp.bin')
+local BAD_PATH = utils.profilename('bad-path-tmp.bin')
+local EXECUTABLE = utils.luacmd(arg)
+
+local function default_payload()
+ -- Preallocate table to avoid table array part reallocations.
+ local _ = table_new(20, 0)
+
+ -- Want too see 20 objects here.
+ for i = 1, 20 do
+ -- Try to avoid crossing with "test" module objects.
+ _[i] = 'memprof-str-'..i
+ end
+
+ _ = nil
+ -- VMSTATE == GC, reported as INTERNAL.
+ collectgarbage()
+end
+
+local function generate_output(filename, payload)
+ -- Clean up all garbage to avoid pollution of free.
+ collectgarbage()
+
+ local res, err = misc.memprof.start(filename)
+ -- Should start succesfully.
+ assert(res, err)
+
+ payload()
+
+ res, err = misc.memprof.stop()
+ -- Should stop succesfully.
+ assert(res, err)
+end
+
+generate_output(TMP_BINFILE, default_payload)
+
+local errcode = os.execute(EXECUTABLE .. ' -tm ' .. BAD_PATH)
+test:ok(errcode ~= 0, 'binfile does not exist')
+
+errcode = os.execute(EXECUTABLE .. ' -tm ' .. TMP_BINFILE)
+test:ok(errcode == 0, 'memprof binfile parsing')
+
+os.remove(TMP_BINFILE)
+os.exit(test:check() and 0 or 1)
diff --git a/tools/CMakeLists.txt b/tools/CMakeLists.txt
index dd7ec6bd..e2e97b63 100644
--- a/tools/CMakeLists.txt
+++ b/tools/CMakeLists.txt
@@ -16,6 +16,7 @@ else()
# path where LuaJIT binary is located.
set(LUAJIT_TOOLS_BIN ${LUAJIT_BINARY_DIR}/${LUAJIT_CLI_NAME})
set(LUAJIT_TOOLS_DIR ${CMAKE_CURRENT_SOURCE_DIR})
+ set(LUAJIT_TOOLS_DIR ${LUAJIT_TOOLS_DIR} PARENT_SCOPE)
# XXX: Unfortunately, there is no convenient way to set
# particular permissions to the output file via CMake.
# Furthermore, I even failed to copy the given file to the same
@@ -77,6 +78,7 @@ else()
"
set(LUAJIT_TOOLS_BIN ${CMAKE_INSTALL_PREFIX}/bin/${LUAJIT_CLI_NAME})
set(LUAJIT_TOOLS_DIR ${CMAKE_INSTALL_PREFIX}/${LUAJIT_DATAROOTDIR})
+ set(LUAJIT_TOOLS_DIR ${LUAJIT_TOOLS_DIR} PARENT_SCOPE)
configure_file(${CMAKE_CURRENT_SOURCE_DIR}/luajit-parse-memprof.in
${PROJECT_BINARY_DIR}/luajit-parse-memprof @ONLY ESCAPE_QUOTES)
file(INSTALL ${PROJECT_BINARY_DIR}/luajit-parse-memprof
--
2.38.1
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [Tarantool-patches] [PATCH luajit v3] memprof: introduce cli flag to run dump parser
2022-12-05 18:01 [Tarantool-patches] [PATCH luajit v3] memprof: introduce cli flag to run dump parser Maxim Kokryashkin via Tarantool-patches
@ 2023-02-07 7:23 ` Sergey Kaplun via Tarantool-patches
0 siblings, 0 replies; 2+ messages in thread
From: Sergey Kaplun via Tarantool-patches @ 2023-02-07 7:23 UTC (permalink / raw)
To: Maxim Kokryashkin; +Cc: tarantool-patches
Hi, Maxim!
Thanks for the fixes!
I belive that this is the last iteration of the review, so LGTM,
after you'll fix some comments below.
On 05.12.22, Maxim Kokryashkin wrote:
> It is really inconvenient to use a standalone shell script to parse
> memprof dump. That is why this commit introduces a CLI flag for tools
> to the LuaJIT, so now it is possible to parse memprof dump
> as simple as:
> ```
> luajit -tm memprof.bin
> ```
>
> Closes tarantool/tarantool#5688
> ---
> Changes in v3:
> - Fixed comments as per review by Sergey
>
> Branch: https://github.com/tarantool/luajit/tree/fckxorg/gh-5688-cli-for-memprof-parse
> PR: https://github.com/tarantool/tarantool/pull/8002
> Issue: https://github.com/tarantool/tarantool/issues/5688
>
> CMakeLists.txt | 9 +--
> Makefile.original | 7 +-
> src/CMakeLists.txt | 5 ++
> src/lj_tools_conf.h.in | 7 ++
> src/luajit.c | 63 ++++++++++++++++--
> .../gh-5688-memprof-cli-flag.test.lua | 64 +++++++++++++++++++
> tools/CMakeLists.txt | 2 +
> 7 files changed, 145 insertions(+), 12 deletions(-)
> create mode 100644 src/lj_tools_conf.h.in
> create mode 100644 test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua
>
> diff --git a/CMakeLists.txt b/CMakeLists.txt
> index c870cce2..97d0d42f 100644
> --- a/CMakeLists.txt
> +++ b/CMakeLists.txt
<snipped>
> diff --git a/Makefile.original b/Makefile.original
> index 0c92df9e..bb0ab73d 100644
> --- a/Makefile.original
> +++ b/Makefile.original
<snipped>
> diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
> index dffc0a4d..17674a41 100644
> --- a/src/CMakeLists.txt
> +++ b/src/CMakeLists.txt
<snipped>
> diff --git a/src/lj_tools_conf.h.in b/src/lj_tools_conf.h.in
> new file mode 100644
> index 00000000..9f9a2e49
> --- /dev/null
> +++ b/src/lj_tools_conf.h.in
<snipped>
> diff --git a/src/luajit.c b/src/luajit.c
> index 1ca24301..bd9ae8f3 100644
> --- a/src/luajit.c
> +++ b/src/luajit.c
<snipped>
>
> +/*
> +** On most Linux distros, it is the default value for the
> +** maximum length of a string passed to `execve`.
> +** However, there is no common value for other OSes, so
> +** the size of 32 default memory pages is adopted.
> +**/
Typo: s<**/><*/>
> +#define MAX_ENV_VAR 32 * 4096
> +
> +static int update_env_var(const char *name, const char *value)
> +{
> + char env_buf[MAX_ENV_VAR] = "";
> + const char *env = getenv(name);
> + if (env == NULL) {
> + return setenv(name, value, 0);
> + } else {
> + strcpy(env_buf, env);
> + return setenv(name, strcat(env_buf, value), 0);
You shold use non-zero value here, if you want to rewrite the enviroment
variable.
| LUA_PATH=";;" src/luajit -tm /tmp/tmp_memprof.bin
| src/luajit: /home/burii/reviews/luajit/cli-flags/tools/memprof.lua:13: module 'utils.bufread' not found:
| no field package.preload['utils.bufread']
| ...
But after the following patch it works fine:
===================================================================
diff --git a/src/luajit.c b/src/luajit.c
index bd9ae8f3..e40a4d30 100644
--- a/src/luajit.c
+++ b/src/luajit.c
@@ -424,7 +424,7 @@ static int update_env_var(const char *name, const char *value)
return setenv(name, value, 0);
} else {
strcpy(env_buf, env);
- return setenv(name, strcat(env_buf, value), 0);
+ return setenv(name, strcat(env_buf, value), 1);
}
}
===================================================================
| LUA_PATH="./?.lua;;" src/luajit -tm /tmp/tmp_memprof.bin
| ALLOCATIONS
| =(command line):1: 174 events +6096 bytes -0 bytes
| ...
Looks like a good testcase to add.
| + return setenv(name, strcat(env_buf, value), 0);
Minor: I suggest to add a check length for total buffer length. (Yes,
it's highly unlikely, that someone has env var >= 128 KB, but we still
can use ENOMEM in such case.)
Feel free to ignore.
> + }
> +}
> +
> /* check that argument has no extra characters at the end */
> #define notail(x) {if ((x)[2] != '\0') return -1;}
>
> @@ -398,6 +436,7 @@ static int dobytecode(lua_State *L, char **argv)
> #define FLAGS_EXEC 4
> #define FLAGS_OPTION 8
> #define FLAGS_NOENV 16
> +#define FLAGS_TOOL 32
>
> static int collectargs(char **argv, int *flags)
> {
> @@ -419,6 +458,12 @@ static int collectargs(char **argv, int *flags)
> notail(argv[i]);
> *flags |= FLAGS_VERSION;
> break;
> + case 't':
> + *flags |= FLAGS_TOOL;
> + if (argv[i][2] == '\0') return -1;
> + if (argv[i + 1] == NULL) return -1;
> + update_env_var("LUA_PATH", TOOLS_PATH);
Should we check the return value of the call here (EINVAL, or ENOMEM)?
If we don't worry about ENOMEM, feel free to ignore.
> + return i + 1;
> case 'e':
> *flags |= FLAGS_EXEC;
> case 'j': /* LuaJIT extension */
> @@ -474,6 +519,10 @@ static int runargs(lua_State *L, char **argv, int argn)
<snipped>
> diff --git a/test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua b/test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua
> new file mode 100644
> index 00000000..ba2d0219
> --- /dev/null
> +++ b/test/tarantool-tests/gh-5688-memprof-cli-flag.test.lua
> @@ -0,0 +1,64 @@
> +local utils = require('utils')
> +
> +-- XXX: The patch is for luajit only, and it doesn't
Typo: s/luajit/LuaJIT/
> +-- work on Tarantool.
> +-- luacheck: no global
Nit: can we move luacheck comment one line below?
(Firstly don't get is it `utils` global)
> +utils.skipcond(
> + (jit.arch ~= 'x86' and jit.arch ~= 'x64') or _TARANTOOL,
> + jit.arch..' architecture is NIY for memprof'
> +)
> +
> +local tap = require('tap')
> +
> +local test = tap.test('gh-5688-memprof-cli-flag')
> +test:plan(2)
> +
> +jit.off()
> +jit.flush()
> +
> +local table_new = require 'table.new'
> +
> +local TMP_BINFILE = utils.profilename('memprofdata.tmp.bin')
> +local BAD_PATH = utils.profilename('bad-path-tmp.bin')
> +local EXECUTABLE = utils.luacmd(arg)
> +
> +local function default_payload()
> + -- Preallocate table to avoid table array part reallocations.
> + local _ = table_new(20, 0)
Minor: Should it be a constant as far as it is used twice?
Also, do we need this monkey business about table size and so on as far
as we don't check exactly values?
> +
> + -- Want too see 20 objects here.
> + for i = 1, 20 do
> + -- Try to avoid crossing with "test" module objects.
> + _[i] = 'memprof-str-'..i
> + end
> +
> + _ = nil
> + -- VMSTATE == GC, reported as INTERNAL.
> + collectgarbage()
> +end
<snipped>
> diff --git a/tools/CMakeLists.txt b/tools/CMakeLists.txt
> index dd7ec6bd..e2e97b63 100644
> --- a/tools/CMakeLists.txt
> +++ b/tools/CMakeLists.txt
<snipped>
> --
> 2.38.1
>
--
Best regards,
Sergey Kaplun
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2023-02-07 7:27 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-12-05 18:01 [Tarantool-patches] [PATCH luajit v3] memprof: introduce cli flag to run dump parser Maxim Kokryashkin via Tarantool-patches
2023-02-07 7:23 ` Sergey Kaplun via Tarantool-patches
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox