pahole: avoid adding same struct structure to two rb trees
Eduard Zingerman
maintainerPATCH 1/1 pahole: avoid adding same struct structure to two rb trees
When pahole is executed in '-F dwarf --sort' mode there are two places where 'struct structure' instance could be added to the rb_tree:
The first is triggered from the following call stack:
print_classes()
structures__add()
__structures__add()
(adds to global pahole.c:structures__tree)
The second is triggered from the following call stack:
print_ordered_classes()
resort_classes()
resort_add()
(adds to local rb_tree instance)
Both places use the same 'struct structure::rb_node' field, so if both code pathes are executed the final state of the 'structures__tree' might be inconsistent.
For example, this could be observed when DEBUG_CHECK_LEAKS build flag is set. Here is the command line snippet that eventually leads to a segfault:
$ for i in $(seq 1 100); do \
echo $i; \
pahole -F dwarf --flat_arrays --sort --jobs vmlinux > /dev/null \
|| break; \
doneGDB shows the following stack trace:
Thread 1 "pahole" received signal SIGSEGV, Segmentation fault.
0x00007ffff7f819ad in __rb_erase_color (node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:134
134 if (parent->rb_left == node)
(gdb) bt
#0 0x00007ffff7f819ad in __rb_erase_color (node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:134 #1 0x00007ffff7f82014 in rb_erase (node=0x7fff21ae5b80, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:275 #2 0x0000555555559c3d in __structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:440 #3 0x0000555555559c70 in structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:448 #4 0x0000555555560bb6 in main (argc=13, argv=0x7fffffffdcd8) at /home/eddy/work/dwarves-fork/pahole.c:3584
This commit modifies resort_classes() to re-use 'structures__tree' and to reset 'rb_node' fields before adding structure instances to the tree for a second time.
Lock/unlock structures_lock to be consistent with structures_add() and structures__delete() code.
--- pahole.c | 41 ++++++++++++++++++++++++++++------------- 1 file changed, 28 insertions(+), 13 deletions(-)
pahole.c +28 -13
@@ -621,9 +621,9 @@ static void print_classes(struct cu *cu) } } -static void __print_ordered_classes (struct rb_root *root) +static void __print_ordered_classes (void ) { -struct rb_node *next =rb_first (root); +struct rb_node *next =rb_first (&structures__tree);while (next) { struct structure *st = rb_entry(next, struct structure, rb_node);
@@ -660,24 +660,39 @@ static void resort_add(struct rb_root *resorted, struct structure *str)rb_insert_color (&str->rb_node, resorted); } -static void resort_classes (struct rb_root *resorted,struct list_head *head) +static void resort_classes (void ) {struct structure *str; -list_for_each_entry (str, head, node) -resort_add (resorted, str); +pthread_mutex_lock (&structures_lock); + +/* The need_resort flag is set by type__compare_members() +* within the following call stack: +* +* print_classes() +* structures__add() +* __structures__add() +* type__compare() +* +* The call to structures__add() registers 'struct structures' +* instances in both 'structures__tree' and 'structures__list'. +* In order to avoid adding same node to the tree twice reset +* both the 'structures__tree' and 'str->rb_node'. +*/ + structures__tree =RB_ROOT ; +list_for_each_entry (str, &structures__list, node) { +bzero (&str->rb_node,sizeof (str->rb_node)); +resort_add (&structures__tree, str); + } + +pthread_mutex_unlock (&structures_lock); }static void print_ordered_classes (void ) { -if (!need_resort) { -__print_ordered_classes (&structures__tree); - }else { -struct rb_root resorted =RB_ROOT ; - -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); - } +if (need_resort) +resort_classes (); +__print_ordered_classes (); }
Signature · 1 line
2.40.1
Em Fri, May 26, 2023 at 02:59:49AM +0300, Eduard Zingerman escreveu:
Quoted 90 lines
When pahole is executed in '-F dwarf --sort' mode there are two places where 'struct structure' instance could be added to the
rb_tree:The first is triggered from the following call stack:
print_classes()
structures__add()
__structures__add()
(adds to global pahole.c:structures__tree)The second is triggered from the following call stack:
print_ordered_classes()
resort_classes()
resort_add()
(adds to localrb_treeinstance)Both places use the same 'struct
structure::rb_node' field, so if both code pathes are executed the final state of the 'structures__tree' might be inconsistent.For example, this could be observed when
DEBUG_CHECK_LEAKSbuild flag is set. Here is the command line snippet that eventually leads to a segfault:$ for i in $(seq 1 100); do \ echo $i; \ pahole -F dwarf --flat_arrays --sort --jobs vmlinux > /dev/null \ || break; \ doneGDB shows the following stack trace:
Thread 1 "pahole" received signal SIGSEGV, Segmentation fault.
0x00007ffff7f819adin__rb_erase_color(node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at/home/eddy/work/dwarves-fork/rbtree.c:134
134 if (parent->rb_left== node)
(gdb) bt#0 0x00007ffff7f819ad in __rb_erase_color (node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:134 #1 0x00007ffff7f82014 in rb_erase (node=0x7fff21ae5b80, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:275 #2 0x0000555555559c3d in __structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:440 #3 0x0000555555559c70 in structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:448 #4 0x0000555555560bb6 in main (argc=13, argv=0x7fffffffdcd8) at /home/eddy/work/dwarves-fork/pahole.c:3584This commit modifies
resort_classes()to re-use 'structures__tree' and to reset 'rb_node' fields before adding structure instances to the tree for a second time.Lock/unlock
structures_lockto be consistent withstructures_add()andstructures__delete()code.Signed-off-by Eduard Zingerman <[email protected]>--- pahole.c | 41 ++++++++++++++++++++++++++++------------- 1 file changed, 28 insertions(+), 13 deletions(-)pahole.c +11
-5@@ -621,9 +621,9 @@ static void print_classes(struct cu *cu) } } -static void __print_ordered_classes (struct rb_root *root) +static void __print_ordered_classes (void ) { -struct rb_node *next =rb_first (root); +struct rb_node *next =rb_first (&structures__tree);while (next) { struct structure *st = rb_entry(next, struct structure, rb_node);@@ -660,24 +660,39 @@ static void resort_add(struct rb_root *resorted, struct structure *str)rb_insert_color (&str->rb_node, resorted); } -static void resort_classes (struct rb_root *resorted,struct list_head *head) +static void resort_classes (void ) {struct structure *str; -list_for_each_entry (str, head, node) -resort_add (resorted, str); +pthread_mutex_lock (&structures_lock); + +/* The need_resort flag is set by type__compare_members() +* within the following call stack: +* +* print_classes() +* structures__add() +* __structures__add()
pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 684
Quoted 10 lines
pahole.c +10
-0+* type__compare() +* +* The call to structures__add() registers 'struct structures' +* instances in both 'structures__tree' and 'structures__list'. +* In order to avoid adding same node to the tree twice reset +* both the 'structures__tree' and 'str->rb_node'. +*/ + structures__tree =RB_ROOT ; +list_for_each_entry (str, &structures__list, node) { +bzero (&str->rb_node,sizeof (str->rb_node));
Why is this bzero needed?
pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 685
Quoted 1 line
pahole.c +1
-0+resort_add (&structures__tree, str);
resort_add will call rb_link_node(&str->rb_node, parent, p); and it, in turn:
static inline void rb_link_node(struct rb_node * node, struct rb_node * parent,
struct rb_node ** rb_link)
{
node->rb_parent_color = (unsigned long )parent;
node->rb_left = node->rb_right = NULL;
*rb_link = node;
}And:
struct rb_node
{
unsigned long rb_parent_color;
#define RB_RED 0
#define RB_BLACK 1
struct rb_node *rb_right;
struct rb_node *rb_left;
} __attribute__((aligned(sizeof(long))))So all the fields are being initialized in the operation right after the bzero(), no?
- Arnaldo
Quoted 24 lines
pahole.c +6
-8+ } + +pthread_mutex_unlock (&structures_lock); }static void print_ordered_classes (void ) { -if (!need_resort) { -__print_ordered_classes (&structures__tree); - }else { -struct rb_root resorted =RB_ROOT ; - -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); - } +if (need_resort) +resort_classes (); +__print_ordered_classes (); }Signature · 1 line
2.40.1
Signature · 1 line
- Arnaldo
Eduard Zingerman
maintainerQuoted 129 lines · Arnaldo Carvalho de Melo
Em Fri, May 26, 2023 at 02:59:49AM +0300, Eduard Zingerman escreveu:
Quoted 90 lines
When pahole is executed in '-F dwarf --sort' mode there are two places where 'struct structure' instance could be added to the
rb_tree:The first is triggered from the following call stack:
print_classes()
structures__add()
__structures__add()
(adds to global pahole.c:structures__tree)The second is triggered from the following call stack:
print_ordered_classes()
resort_classes()
resort_add()
(adds to localrb_treeinstance)Both places use the same 'struct
structure::rb_node' field, so if both code pathes are executed the final state of the 'structures__tree' might be inconsistent.For example, this could be observed when
DEBUG_CHECK_LEAKSbuild flag is set. Here is the command line snippet that eventually leads to a segfault:$ for i in $(seq 1 100); do \ echo $i; \ pahole -F dwarf --flat_arrays --sort --jobs vmlinux > /dev/null \ || break; \ doneGDB shows the following stack trace:
Thread 1 "pahole" received signal SIGSEGV, Segmentation fault.
0x00007ffff7f819adin__rb_erase_color(node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at/home/eddy/work/dwarves-fork/rbtree.c:134
134 if (parent->rb_left== node)
(gdb) bt#0 0x00007ffff7f819ad in __rb_erase_color (node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:134 #1 0x00007ffff7f82014 in rb_erase (node=0x7fff21ae5b80, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:275 #2 0x0000555555559c3d in __structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:440 #3 0x0000555555559c70 in structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:448 #4 0x0000555555560bb6 in main (argc=13, argv=0x7fffffffdcd8) at /home/eddy/work/dwarves-fork/pahole.c:3584This commit modifies
resort_classes()to re-use 'structures__tree' and to reset 'rb_node' fields before adding structure instances to the tree for a second time.Lock/unlock
structures_lockto be consistent withstructures_add()andstructures__delete()code.Signed-off-by Eduard Zingerman <[email protected]>--- pahole.c | 41 ++++++++++++++++++++++++++++------------- 1 file changed, 28 insertions(+), 13 deletions(-)pahole.c +11
-5@@ -621,9 +621,9 @@ static void print_classes(struct cu *cu) } } -static void __print_ordered_classes (struct rb_root *root) +static void __print_ordered_classes (void ) { -struct rb_node *next =rb_first (root); +struct rb_node *next =rb_first (&structures__tree);while (next) { struct structure *st = rb_entry(next, struct structure, rb_node);@@ -660,24 +660,39 @@ static void resort_add(struct rb_root *resorted, struct structure *str)rb_insert_color (&str->rb_node, resorted); } -static void resort_classes (struct rb_root *resorted,struct list_head *head) +static void resort_classes (void ) {struct structure *str; -list_for_each_entry (str, head, node) -resort_add (resorted, str); +pthread_mutex_lock (&structures_lock); + +/* The need_resort flag is set by type__compare_members() +* within the following call stack: +* +* print_classes() +* structures__add() +* __structures__add() pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 684
Quoted 10 lines
pahole.c +10
-0+* type__compare() +* +* The call to structures__add() registers 'struct structures' +* instances in both 'structures__tree' and 'structures__list'. +* In order to avoid adding same node to the tree twice reset +* both the 'structures__tree' and 'str->rb_node'. +*/ + structures__tree =RB_ROOT ; +list_for_each_entry (str, &structures__list, node) { +bzero (&str->rb_node,sizeof (str->rb_node));Why is this bzero needed?
pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 685
Quoted 1 line
pahole.c +1
-0+resort_add (&structures__tree, str);
resort_addwill callrb_link_node(&str->rb_node, parent, p); and it, in turn:static inline void rb_link_node(struct
rb_node* node, structrb_node* parent,
structrb_node**rb_link){ node->rb_parent_color = (unsigned long )parent; node->rb_left = node->rb_right = NULL; *rb_link = node; }And:
struct
rb_node{ unsigned long rb_parent_color; #define RB_RED 0 #define RB_BLACK 1 struct rb_node *rb_right; struct rb_node *rb_left; } __attribute__((aligned(sizeof(long))))
Quoted 2 lines
So all the fields are being initialized in the operation right after the
bzero(), no?
Right, you are correct.
The 'structures__tree = RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/o bzero().
Quoted 28 lines
- Arnaldo
Quoted 24 lines
pahole.c +6
-8+ } + +pthread_mutex_unlock (&structures_lock); }static void print_ordered_classes (void ) { -if (!need_resort) { -__print_ordered_classes (&structures__tree); - }else { -struct rb_root resorted =RB_ROOT ; - -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); - } +if (need_resort) +resort_classes (); +__print_ordered_classes (); }Signature · 1 line
2.40.1
Em Fri, Jun 02, 2023 at 04:52:40PM +0300, Eduard Zingerman escreveu:
Quoted 130 lines · Arnaldo Carvalho de Melo
Quoted 129 lines
Em Fri, May 26, 2023 at 02:59:49AM +0300, Eduard Zingerman escreveu:
Quoted 90 lines
When pahole is executed in '-F dwarf --sort' mode there are two places where 'struct structure' instance could be added to the
rb_tree:The first is triggered from the following call stack:
print_classes()
structures__add()
__structures__add()
(adds to global pahole.c:structures__tree)The second is triggered from the following call stack:
print_ordered_classes()
resort_classes()
resort_add()
(adds to localrb_treeinstance)Both places use the same 'struct
structure::rb_node' field, so if both code pathes are executed the final state of the 'structures__tree' might be inconsistent.For example, this could be observed when
DEBUG_CHECK_LEAKSbuild flag is set. Here is the command line snippet that eventually leads to a segfault:$ for i in $(seq 1 100); do \ echo $i; \ pahole -F dwarf --flat_arrays --sort --jobs vmlinux > /dev/null \ || break; \ doneGDB shows the following stack trace:
Thread 1 "pahole" received signal SIGSEGV, Segmentation fault.
0x00007ffff7f819adin__rb_erase_color(node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at/home/eddy/work/dwarves-fork/rbtree.c:134
134 if (parent->rb_left== node)
(gdb) bt#0 0x00007ffff7f819ad in __rb_erase_color (node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:134 #1 0x00007ffff7f82014 in rb_erase (node=0x7fff21ae5b80, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:275 #2 0x0000555555559c3d in __structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:440 #3 0x0000555555559c70 in structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:448 #4 0x0000555555560bb6 in main (argc=13, argv=0x7fffffffdcd8) at /home/eddy/work/dwarves-fork/pahole.c:3584This commit modifies
resort_classes()to re-use 'structures__tree' and to reset 'rb_node' fields before adding structure instances to the tree for a second time.Lock/unlock
structures_lockto be consistent withstructures_add()andstructures__delete()code.Signed-off-by Eduard Zingerman <[email protected]>--- pahole.c | 41 ++++++++++++++++++++++++++++------------- 1 file changed, 28 insertions(+), 13 deletions(-)pahole.c +11
-5@@ -621,9 +621,9 @@ static void print_classes(struct cu *cu) } } -static void __print_ordered_classes (struct rb_root *root) +static void __print_ordered_classes (void ) { -struct rb_node *next =rb_first (root); +struct rb_node *next =rb_first (&structures__tree);while (next) { struct structure *st = rb_entry(next, struct structure, rb_node);@@ -660,24 +660,39 @@ static void resort_add(struct rb_root *resorted, struct structure *str)rb_insert_color (&str->rb_node, resorted); } -static void resort_classes (struct rb_root *resorted,struct list_head *head) +static void resort_classes (void ) {struct structure *str; -list_for_each_entry (str, head, node) -resort_add (resorted, str); +pthread_mutex_lock (&structures_lock); + +/* The need_resort flag is set by type__compare_members() +* within the following call stack: +* +* print_classes() +* structures__add() +* __structures__add() pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 684
Quoted 10 lines
pahole.c +10
-0+* type__compare() +* +* The call to structures__add() registers 'struct structures' +* instances in both 'structures__tree' and 'structures__list'. +* In order to avoid adding same node to the tree twice reset +* both the 'structures__tree' and 'str->rb_node'. +*/ + structures__tree =RB_ROOT ; +list_for_each_entry (str, &structures__list, node) { +bzero (&str->rb_node,sizeof (str->rb_node));Why is this bzero needed?
pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 685
Quoted 1 line
pahole.c +1
-0+resort_add (&structures__tree, str);
resort_addwill callrb_link_node(&str->rb_node, parent, p); and it, in turn:static inline void rb_link_node(struct
rb_node* node, structrb_node* parent,
structrb_node**rb_link){ node->rb_parent_color = (unsigned long )parent; node->rb_left = node->rb_right = NULL; *rb_link = node; }And:
struct
rb_node{ unsigned long rb_parent_color; #define RB_RED 0 #define RB_BLACK 1 struct rb_node *rb_right; struct rb_node *rb_left; } __attribute__((aligned(sizeof(long))))
Quoted 6 lines
Quoted 2 lines
So all the fields are being initialized in the operation right after the
bzero(), no?Right, you are correct.
The 'structures__tree=RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/obzero().
Humm, so basically this boils down to the following patch?
- Arnaldo
pahole.c +6 -1
@@ -674,7 +674,12 @@ static void print_ordered_classes(void)__print_ordered_classes (&structures__tree); }else {struct rb_root resorted =RB_ROOT ; - +#ifdef DEBUG_CHECK_LEAKS +// We'll delete structures from structures__tree, since we're +// adding them to ther resorted list, better not keep +// references there. + structures__tree =RB_ROOT ; +#endif resort_classes (&resorted, &structures__list);__print_ordered_classes (&resorted); }
Eduard Zingerman
maintainerQuoted 147 lines · Arnaldo Carvalho de Melo
Em Fri, Jun 02, 2023 at 04:52:40PM +0300, Eduard Zingerman escreveu:
Quoted 130 lines · Arnaldo Carvalho de Melo
Quoted 129 lines
Em Fri, May 26, 2023 at 02:59:49AM +0300, Eduard Zingerman escreveu:
Quoted 90 lines
When pahole is executed in '-F dwarf --sort' mode there are two places where 'struct structure' instance could be added to the
rb_tree:The first is triggered from the following call stack:
print_classes()
structures__add()
__structures__add()
(adds to global pahole.c:structures__tree)The second is triggered from the following call stack:
print_ordered_classes()
resort_classes()
resort_add()
(adds to localrb_treeinstance)Both places use the same 'struct
structure::rb_node' field, so if both code pathes are executed the final state of the 'structures__tree' might be inconsistent.For example, this could be observed when
DEBUG_CHECK_LEAKSbuild flag is set. Here is the command line snippet that eventually leads to a segfault:$ for i in $(seq 1 100); do \ echo $i; \ pahole -F dwarf --flat_arrays --sort --jobs vmlinux > /dev/null \ || break; \ doneGDB shows the following stack trace:
Thread 1 "pahole" received signal SIGSEGV, Segmentation fault.
0x00007ffff7f819adin__rb_erase_color(node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at/home/eddy/work/dwarves-fork/rbtree.c:134
134 if (parent->rb_left== node)
(gdb) bt#0 0x00007ffff7f819ad in __rb_erase_color (node=0x7fffd4045830, parent=0x0, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:134 #1 0x00007ffff7f82014 in rb_erase (node=0x7fff21ae5b80, root=0x5555555672d8 <structures.tree>) at /home/eddy/work/dwarves-fork/rbtree.c:275 #2 0x0000555555559c3d in __structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:440 #3 0x0000555555559c70 in structures__delete () at /home/eddy/work/dwarves-fork/pahole.c:448 #4 0x0000555555560bb6 in main (argc=13, argv=0x7fffffffdcd8) at /home/eddy/work/dwarves-fork/pahole.c:3584This commit modifies
resort_classes()to re-use 'structures__tree' and to reset 'rb_node' fields before adding structure instances to the tree for a second time.Lock/unlock
structures_lockto be consistent withstructures_add()andstructures__delete()code.Signed-off-by Eduard Zingerman <[email protected]>--- pahole.c | 41 ++++++++++++++++++++++++++++------------- 1 file changed, 28 insertions(+), 13 deletions(-)pahole.c +11
-5@@ -621,9 +621,9 @@ static void print_classes(struct cu *cu) } } -static void __print_ordered_classes (struct rb_root *root) +static void __print_ordered_classes (void ) { -struct rb_node *next =rb_first (root); +struct rb_node *next =rb_first (&structures__tree);while (next) { struct structure *st = rb_entry(next, struct structure, rb_node);@@ -660,24 +660,39 @@ static void resort_add(struct rb_root *resorted, struct structure *str)rb_insert_color (&str->rb_node, resorted); } -static void resort_classes (struct rb_root *resorted,struct list_head *head) +static void resort_classes (void ) {struct structure *str; -list_for_each_entry (str, head, node) -resort_add (resorted, str); +pthread_mutex_lock (&structures_lock); + +/* The need_resort flag is set by type__compare_members() +* within the following call stack: +* +* print_classes() +* structures__add() +* __structures__add() pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 684
Quoted 10 lines
pahole.c +10
-0+* type__compare() +* +* The call to structures__add() registers 'struct structures' +* instances in both 'structures__tree' and 'structures__list'. +* In order to avoid adding same node to the tree twice reset +* both the 'structures__tree' and 'str->rb_node'. +*/ + structures__tree =RB_ROOT ; +list_for_each_entry (str, &structures__list, node) { +bzero (&str->rb_node,sizeof (str->rb_node));Why is this bzero needed?
pahole.c static void resort_add(struct rb_root *resorted, struct structure *str) · line 685
Quoted 1 line
pahole.c +1
-0+resort_add (&structures__tree, str);
resort_addwill callrb_link_node(&str->rb_node, parent, p); and it, in turn:static inline void rb_link_node(struct
rb_node* node, structrb_node* parent,
structrb_node**rb_link){ node->rb_parent_color = (unsigned long )parent; node->rb_left = node->rb_right = NULL; *rb_link = node; }And:
struct
rb_node{ unsigned long rb_parent_color; #define RB_RED 0 #define RB_BLACK 1 struct rb_node *rb_right; struct rb_node *rb_left; } __attribute__((aligned(sizeof(long))))Quoted 6 lines
Quoted 2 lines
So all the fields are being initialized in the operation right after the
bzero(), no?Right, you are correct.
The 'structures__tree=RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/obzero().Humm, so basically this boils down to the following patch?
- Arnaldo
pahole.c +0
-0
pahole.c static void print_ordered_classes(void) · line 682
Quoted 10 lines
pahole.c +6
-1@@ -674,7 +674,12 @@ static void print_ordered_classes(void)__print_ordered_classes (&structures__tree); }else {struct rb_root resorted =RB_ROOT ; - +#ifdef DEBUG_CHECK_LEAKS +// We'll delete structures from structures__tree, since we're +// adding them to ther resorted list, better not keep +// references there. + structures__tree =RB_ROOT ; +#endif
But __structures__delete iterates over structures__tree, so it won't delete anything if code like this, right?
Quoted 3 lines
pahole.c +0
-0resort_classes (&resorted, &structures__list);__print_ordered_classes (&resorted); }
Em Fri, Jun 02, 2023 at 09:08:51PM +0300, Eduard Zingerman escreveu:
Quoted 8 lines · Arnaldo Carvalho de Melo
Quoted 6 lines
Em Fri, Jun 02, 2023 at 04:52:40PM +0300, Eduard Zingerman escreveu:
Quoted 3 lines
Right, you are correct.
The 'structures__tree=RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/obzero().Humm, so basically this boils down to the following patch?
Quoted 19 lines
Review static void print_ordered_classes(void) · line 682
Quoted 12 lines
+++
b/pahole.c(quoted diff) +6
-1@@ -674,7 +674,12 @@ static void print_ordered_classes(void) __print_ordered_classes(&structures__tree); } else { struct rb_root resorted = RB_ROOT; - +#ifdef DEBUG_CHECK_LEAKS + // We'll delete structures from structures__tree, since we're + // adding them to ther resorted list, better not keep + // references there. + structures__tree = RB_ROOT; +#endifBut
__structures__deleteiterates overstructures__tree, so it won't delete anything if code like this, right?Quoted 3 lines
(quoted diff) +0
-0resort_classes(&resorted, &structures__list); __print_ordered_classes(&resorted); }
Yeah, I tried to be minimalistic, my version avoids the crash, but defeats the DEBUG_CHECK_LEAKS purpose :-\
How about:
pahole.c +3 -3
@@ -673,10 +673,10 @@ static void print_ordered_classes(void)if (!need_resort) {__print_ordered_classes (&structures__tree); }else { -struct rb_root resorted =RB_ROOT ; + structures__tree =RB_ROOT ; -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); +resort_classes (&structures__tree, &structures__list); +__print_ordered_classes (&structures__tree); } }
Eduard Zingerman
maintainerQuoted 42 lines · Arnaldo Carvalho de Melo
Em Fri, Jun 02, 2023 at 09:08:51PM +0300, Eduard Zingerman escreveu:
Quoted 8 lines · Arnaldo Carvalho de Melo
Quoted 6 lines
Em Fri, Jun 02, 2023 at 04:52:40PM +0300, Eduard Zingerman escreveu:
Quoted 3 lines
Right, you are correct.
The 'structures__tree=RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/obzero().Humm, so basically this boils down to the following patch?
Quoted 19 lines
Review static void print_ordered_classes(void) · line 682
Quoted 12 lines
+++
b/pahole.c(quoted diff) +6
-1@@ -674,7 +674,12 @@ static void print_ordered_classes(void) __print_ordered_classes(&structures__tree); } else { struct rb_root resorted = RB_ROOT; - +#ifdef DEBUG_CHECK_LEAKS + // We'll delete structures from structures__tree, since we're + // adding them to ther resorted list, better not keep + // references there. + structures__tree = RB_ROOT; +#endifBut
__structures__deleteiterates overstructures__tree, so it won't delete anything if code like this, right?Quoted 3 lines
(quoted diff) +0
-0resort_classes(&resorted, &structures__list); __print_ordered_classes(&resorted); }Yeah, I tried to be minimalistic, my version avoids the crash, but defeats the
DEBUG_CHECK_LEAKSpurpose :-\How about:
pahole.c +0
-0@@ -673,10 +673,10 @@ static void print_ordered_classes(void)if (!need_resort) {__print_ordered_classes (&structures__tree); }else {
pahole.c static void print_ordered_classes(void) · line 682
Quoted 10 lines
pahole.c +3
-3-struct rb_root resorted =RB_ROOT ; + structures__tree =RB_ROOT ; -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); +resort_classes (&structures__tree, &structures__list); +__print_ordered_classes (&structures__tree); } }
That would work, but I still think that there is no need to replicate call to __print_ordered_classes, as long as the same list is passed as an argument, e.g.:
@@ -670,14 +671,11 @@ static void resort_classes(struct rb_root *resorted, struct list_head *head)
static void print_ordered_classes(void)
{
- if (!need_resort) {
- __print_ordered_classes(&structures__tree);
- } else {
- struct rb_root resorted = RB_ROOT;
-
- resort_classes(&resorted, &structures__list);
- __print_ordered_classes(&resorted);
+ if (need_resort) {
+ structures__tree = RB_ROOT;
+ resort_classes(&structures__tree, &structures__list);
}+ __print_ordered_classes(&structures__tree);
}
Em Mon, Jun 05, 2023 at 05:39:19PM +0300, Eduard Zingerman escreveu:
Quoted 65 lines · Arnaldo Carvalho de Melo
Quoted 42 lines
Em Fri, Jun 02, 2023 at 09:08:51PM +0300, Eduard Zingerman escreveu:
Quoted 8 lines · Arnaldo Carvalho de Melo
Quoted 6 lines
Em Fri, Jun 02, 2023 at 04:52:40PM +0300, Eduard Zingerman escreveu:
Quoted 3 lines
Right, you are correct.
The 'structures__tree=RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/obzero().Humm, so basically this boils down to the following patch?
Quoted 19 lines
Review static void print_ordered_classes(void) · line 682
Quoted 12 lines
+++
b/pahole.c(quoted diff) +6
-1@@ -674,7 +674,12 @@ static void print_ordered_classes(void) __print_ordered_classes(&structures__tree); } else { struct rb_root resorted = RB_ROOT; - +#ifdef DEBUG_CHECK_LEAKS + // We'll delete structures from structures__tree, since we're + // adding them to ther resorted list, better not keep + // references there. + structures__tree = RB_ROOT; +#endifBut
__structures__deleteiterates overstructures__tree, so it won't delete anything if code like this, right?Quoted 3 lines
(quoted diff) +0
-0resort_classes(&resorted, &structures__list); __print_ordered_classes(&resorted); }Yeah, I tried to be minimalistic, my version avoids the crash, but defeats the
DEBUG_CHECK_LEAKSpurpose :-\How about:
pahole.c +0
-0@@ -673,10 +673,10 @@ static void print_ordered_classes(void)if (!need_resort) {__print_ordered_classes (&structures__tree); }else {pahole.c static void print_ordered_classes(void) · line 682
Quoted 10 lines
pahole.c +3
-3-struct rb_root resorted =RB_ROOT ; + structures__tree =RB_ROOT ; -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); +resort_classes (&structures__tree, &structures__list); +__print_ordered_classes (&structures__tree); } }That would work, but I still think that there is no need to replicate call to
__print_ordered_classes, as long as the same list is passed as an argument, e.g.:(quoted diff) +0
-3@@ -670,14 +671,11 @@ static void resort_classes(struct rb_root *resorted, struct list_head *head) static void print_ordered_classes(void) { - if (!need_resort) { - __print_ordered_classes(&structures__tree); - } else {
Review static void resort_classes(struct rb_root *resorted, struct list_head *head) · line 679
Quoted 10 lines
(quoted diff) +4
-4- struct rb_root resorted = RB_ROOT; - - resort_classes(&resorted, &structures__list); - __print_ordered_classes(&resorted); + if (need_resort) { + structures__tree = RB_ROOT; + resort_classes(&structures__tree, &structures__list); } + __print_ordered_classes(&structures__tree); }
Right, that can be done as a follow up patch, further simplifying the code.
I'm just trying to have each patch as small as possible.
- Arnaldo
Em Mon, Jun 05, 2023 at 03:54:30PM -0300, Arnaldo Carvalho de Melo escreveu:
Quoted 56 lines
Em Mon, Jun 05, 2023 at 05:39:19PM +0300, Eduard Zingerman escreveu:
Quoted 55 lines · Arnaldo Carvalho de Melo
Quoted 42 lines
Em Fri, Jun 02, 2023 at 09:08:51PM +0300, Eduard Zingerman escreveu:
Quoted 8 lines · Arnaldo Carvalho de Melo
Quoted 6 lines
Em Fri, Jun 02, 2023 at 04:52:40PM +0300, Eduard Zingerman escreveu:
Quoted 3 lines
Right, you are correct.
The 'structures__tree=RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/obzero().Humm, so basically this boils down to the following patch?
Quoted 19 lines
Review static void print_ordered_classes(void) · line 682
Quoted 12 lines
+++
b/pahole.c(quoted diff) +6
-1@@ -674,7 +674,12 @@ static void print_ordered_classes(void) __print_ordered_classes(&structures__tree); } else { struct rb_root resorted = RB_ROOT; - +#ifdef DEBUG_CHECK_LEAKS + // We'll delete structures from structures__tree, since we're + // adding them to ther resorted list, better not keep + // references there. + structures__tree = RB_ROOT; +#endifBut
__structures__deleteiterates overstructures__tree, so it won't delete anything if code like this, right?Quoted 3 lines
(quoted diff) +0
-0resort_classes(&resorted, &structures__list); __print_ordered_classes(&resorted); }Yeah, I tried to be minimalistic, my version avoids the crash, but defeats the
DEBUG_CHECK_LEAKSpurpose :-\How about:
pahole.c +0
-0@@ -673,10 +673,10 @@ static void print_ordered_classes(void)if (!need_resort) {__print_ordered_classes (&structures__tree); }else {pahole.c static void print_ordered_classes(void) · line 682
Quoted 10 lines
pahole.c +3
-3-struct rb_root resorted =RB_ROOT ; + structures__tree =RB_ROOT ; -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); +resort_classes (&structures__tree, &structures__list); +__print_ordered_classes (&structures__tree); } }That would work, but I still think that there is no need to replicate call
I'm going thru the pile of stuff from before my vacations, can I take the above as an Acked-by in addition to your Reported-by?
- Arnaldo
Quoted 25 lines
Quoted 10 lines
to
__print_ordered_classes, as long as the same list is passed as an argument, e.g.:(quoted diff) +0
-3@@ -670,14 +671,11 @@ static void resort_classes(struct rb_root *resorted, struct list_head *head) static void print_ordered_classes(void) { - if (!need_resort) { - __print_ordered_classes(&structures__tree); - } else {Review static void resort_classes(struct rb_root *resorted, struct list_head *head) · line 679
Quoted 10 lines
(quoted diff) +4
-4- struct rb_root resorted = RB_ROOT; - - resort_classes(&resorted, &structures__list); - __print_ordered_classes(&resorted); + if (need_resort) { + structures__tree = RB_ROOT; + resort_classes(&structures__tree, &structures__list); } + __print_ordered_classes(&structures__tree); }Right, that can be done as a follow up patch, further simplifying the code.
I'm just trying to have each patch as small as possible.
Eduard Zingerman
maintainerQuoted 58 lines · Arnaldo Carvalho de Melo
Em Mon, Jun 05, 2023 at 03:54:30PM -0300, Arnaldo Carvalho de Melo escreveu:
Quoted 56 lines
Em Mon, Jun 05, 2023 at 05:39:19PM +0300, Eduard Zingerman escreveu:
Quoted 55 lines · Arnaldo Carvalho de Melo
Quoted 42 lines
Em Fri, Jun 02, 2023 at 09:08:51PM +0300, Eduard Zingerman escreveu:
Quoted 8 lines · Arnaldo Carvalho de Melo
Quoted 6 lines
Em Fri, Jun 02, 2023 at 04:52:40PM +0300, Eduard Zingerman escreveu:
Quoted 3 lines
Right, you are correct.
The 'structures__tree=RB_ROOT' part is still necessary, though. If you are ok with overall structure of the patch I can resend it w/obzero().Humm, so basically this boils down to the following patch?
Quoted 19 lines
Review static void print_ordered_classes(void) · line 682
Quoted 12 lines
+++
b/pahole.c(quoted diff) +6
-1@@ -674,7 +674,12 @@ static void print_ordered_classes(void) __print_ordered_classes(&structures__tree); } else { struct rb_root resorted = RB_ROOT; - +#ifdef DEBUG_CHECK_LEAKS + // We'll delete structures from structures__tree, since we're + // adding them to ther resorted list, better not keep + // references there. + structures__tree = RB_ROOT; +#endifBut
__structures__deleteiterates overstructures__tree, so it won't delete anything if code like this, right?Quoted 3 lines
(quoted diff) +0
-0resort_classes(&resorted, &structures__list); __print_ordered_classes(&resorted); }Yeah, I tried to be minimalistic, my version avoids the crash, but defeats the
DEBUG_CHECK_LEAKSpurpose :-\How about:
pahole.c +0
-0@@ -673,10 +673,10 @@ static void print_ordered_classes(void)if (!need_resort) {__print_ordered_classes (&structures__tree); }else {pahole.c static void print_ordered_classes(void) · line 682
Quoted 10 lines
pahole.c +3
-3-struct rb_root resorted =RB_ROOT ; + structures__tree =RB_ROOT ; -resort_classes (&resorted, &structures__list); -__print_ordered_classes (&resorted); +resort_classes (&structures__tree, &structures__list); +__print_ordered_classes (&structures__tree); } }That would work, but I still think that there is no need to replicate call
Quoted 2 lines
I'm going thru the pile of stuff from before my vacations, can I take the above as an Acked-by in addition to your Reported-by?
Hi Arnaldo,
Sure, no problem.
Quoted 28 lines
- Arnaldo
Quoted 25 lines
Quoted 10 lines
to
__print_ordered_classes, as long as the same list is passed as an argument, e.g.:(quoted diff) +0
-3@@ -670,14 +671,11 @@ static void resort_classes(struct rb_root *resorted, struct list_head *head) static void print_ordered_classes(void) { - if (!need_resort) { - __print_ordered_classes(&structures__tree); - } else {Review static void resort_classes(struct rb_root *resorted, struct list_head *head) · line 679
Quoted 10 lines
(quoted diff) +4
-4- struct rb_root resorted = RB_ROOT; - - resort_classes(&resorted, &structures__list); - __print_ordered_classes(&resorted); + if (need_resort) { + structures__tree = RB_ROOT; + resort_classes(&structures__tree, &structures__list); } + __print_ordered_classes(&structures__tree); }Right, that can be done as a follow up patch, further simplifying the code.
I'm just trying to have each patch as small as possible.