Skip to content

GH-123044: Give the POP_TOP after a case test a location in the body, not the pattern. #130627

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 10 commits into from
Mar 10, 2025
Merged

GH-123044: Give the POP_TOP after a case test a location in the body, not the pattern. #130627

Show file tree
Hide file tree
Changes from 6 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
fd5d1e7
Use case body location for POP_TOP after case test, to give better lo…
markshannon Feb 27, 2025
1ce7f7f
Add news
markshannon Feb 27, 2025
f98fe7a
Include a default case in test
markshannon Feb 27, 2025
7084af0
Fix copy and paste
markshannon Feb 27, 2025
9332b19
Fix formatting in NEWS file for match cases
markshannon Feb 27, 2025
0d82f1b
Use pseudo location to mark instruction as having the location of the…
markshannon Feb 28, 2025
ab6fa4c
Use -2s for NEXT_LOCATION. Modify basicblock_remove_redundant_nops an…
markshannon Mar 6, 2025
1229037
Add a couple of asserts for NEXT_LOCATION
markshannon Mar 6, 2025
a077433
Replace NEXT_LOCATION with NO_LOCATION if there is no meaningful loca…
markshannon Mar 7, 2025
00380b3
Merge branch 'main' into location-pop-top-after-case
markshannon Mar 10, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions Include/internal/pycore_symtable.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,7 @@ typedef struct {
.end_col_offset = (n)->end_col_offset }

static const _Py_SourceLocation NO_LOCATION = {-1, -1, -1, -1};
static const _Py_SourceLocation NEXT_LOCATION = {INT_MAX, INT_MAX, INT_MAX, INT_MAX};
Copy link
Member

@iritkatriel iritkatriel Mar 1, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is a good idea.

We probably need to change remove_redundant_nops to treat this as no location. Maybe it should be (-2, -2, -2, -2) and then just compare to 0?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are some checks for NO_LOCATION that compare to 0, but it isn't clear to me which should support NEXT_LOCATION and which shouldn't, so I'm a bit reluctant to do that.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

With {2, -2, -2, -2}, basicblock_remove_redundant_nops and propagate_line_numbers needed changing.
basicblock_remove_redundant_nops needs to treat NEXT_LOCATION like NO_LOCATION
propagate_line_numbers needs to treat NEXT_LOCATION not like NO_LOCATION


/* __future__ information */
typedef struct {
Expand Down
29 changes: 29 additions & 0 deletions Lib/test/test_monitoring.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1681,6 +1681,35 @@ async def foo():
('branch left', 'func', 12, 12)])


def test_match(self):

def func(v=1):
x = 0
for v in range(4):
match v:
case 1:
x += 1
case 2:
x += 2
case _:
x += 3
return x

self.check_events(func, recorders = BRANCHES_RECORDERS, expected = [
('branch left', 'func', 2, 2),
('branch right', 'func', 4, 6),
('branch right', 'func', 6, 8),
('branch left', 'func', 2, 2),
('branch left', 'func', 4, 5),
('branch left', 'func', 2, 2),
('branch right', 'func', 4, 6),
('branch left', 'func', 6, 7),
('branch left', 'func', 2, 2),
('branch right', 'func', 4, 6),
('branch right', 'func', 6, 8),
('branch right', 'func', 2, 10)])


class TestBranchConsistency(MonitoringTestBase, unittest.TestCase):

def check_branches(self, func, tool=TEST_TOOL, recorders=BRANCH_OFFSET_RECORDERS):
Expand Down
2 changes: 2 additions & 0 deletions Misc/NEWS.d/next/Core_and_Builtins/2025-02-27-10-47-09.gh-issue-123044.8182Un.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
Make sure that the location of branch targets in ``match`` cases is in the
body, not the pattern.
9 changes: 9 additions & 0 deletions Python/assemble.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -343,6 +343,15 @@ assemble_location_info(struct assembler *a, instr_sequence *instrs,
a->a_lineno = firstlineno;
location loc = NO_LOCATION;
int size = 0;
if (same_location(instrs->s_instrs[instrs->s_used-1].i_loc, NEXT_LOCATION)) {
instrs->s_instrs[instrs->s_used-1].i_loc = NO_LOCATION;
}
for (int i = instrs->s_used-1; i > 0; i--) {
instruction *instr = &instrs->s_instrs[i];
if (same_location(instr[-1].i_loc, NEXT_LOCATION)) {
instr[-1].i_loc = instr->i_loc;
}
}
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this transformation should be in propagate_line_numbers in flowgraph.c. Otherwise the last instruction may remain without location, when it could have received a location from an earlier instruction in its block.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wouldn't it be too early in propagate_line_numbers?
If the NEXT_LOCATION instruction is at the end of the block where would it get the location information from?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

From the next block (fall through or jump). If there is more than one successor, then we have a problem anyway, right, so what do we do?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should only use NEXT_LOCATION if instruction is not the last instruction in the block. I'll add an assert for that.

for (int i = 0; i < instrs->s_used; i++) {
instruction *instr = &instrs->s_instrs[i];
if (!same_location(loc, instr->i_loc)) {
Expand Down
3 changes: 2 additions & 1 deletion Python/codegen.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6121,7 +6121,8 @@ codegen_match_inner(compiler *c, stmt_ty s, pattern_context *pc)
}
// Success! Pop the subject off, we're done with it:
if (i != cases - has_default - 1) {
ADDOP(c, LOC(m->pattern), POP_TOP);
/* Use the next location to give better locations for branch events */
ADDOP(c, NEXT_LOCATION, POP_TOP);
}
VISIT_SEQ(c, stmt, m->body);
ADDOP_JUMP(c, NO_LOCATION, JUMP, end);
Expand Down
Loading