[xen master] x86/cpuid: Calculate FEATURESET_NR_ENTRIES more helpfully

[patchbot@xxxxxxx]

commit 56e2c8e5860090a35d5f0cafe168223a2a7c0e62
Author:     Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
AuthorDate: Wed May 10 19:58:43 2023 +0100
Commit:     Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
CommitDate: Mon May 15 11:03:11 2023 +0100

    x86/cpuid: Calculate FEATURESET_NR_ENTRIES more helpfully
    
    When adding new featureset words, it is convenient to split the work into
    several patches.  However, GCC 12 spotted that the way we prefer to split 
the
    work results in a real (transient) breakage whereby the policy <-> 
featureset
    helpers perform out-of-bounds accesses on the featureset array.
    
    Fix this by having gen-cpuid.py calculate FEATURESET_NR_ENTRIES from the
    comments describing the word blocks, rather than from the XEN_CPUFEATURE()
    with the greatest value.
    
    For simplicty, require that the word blocks appear in order.  This can be
    revisted if we find a good reason to have blocks out of order.
    
    No functional change.
    
    Reported-by: Jan Beulich <jbeulich@xxxxxxxx>
    Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
    Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>
---
 xen/tools/gen-cpuid.py | 42 ++++++++++++++++++++++++++++++++++++------
 1 file changed, 36 insertions(+), 6 deletions(-)

diff --git a/xen/tools/gen-cpuid.py b/xen/tools/gen-cpuid.py
index 0edb7d4a19..86d00bb3c2 100755
--- a/xen/tools/gen-cpuid.py
+++ b/xen/tools/gen-cpuid.py
@@ -50,13 +50,37 @@ def parse_definitions(state):
         "\s+([\s\d]+\*[\s\d]+\+[\s\d]+)\)"
         "\s+/\*([\w!]*) .*$")
 
+    word_regex = re.compile(
+        r"^/\* .* word (\d*) \*/$")
+    last_word = -1
+
     this = sys.modules[__name__]
 
     for l in state.input.readlines():
-        # Short circuit the regex...
-        if not l.startswith("XEN_CPUFEATURE("):
+
+        # Short circuit the regexes...
+        if not (l.startswith("XEN_CPUFEATURE(") or
+                l.startswith("/* ")):
             continue
 
+        # Handle /* ... word $N */ lines
+        if l.startswith("/* "):
+
+            res = word_regex.match(l)
+            if res is None:
+                continue # Some other comment
+
+            word = int(res.groups()[0])
+
+            if word != last_word + 1:
+                raise Fail("Featureset word %u out of order (last word %u)"
+                           % (word, last_word))
+
+            last_word = word
+            state.nr_entries = word + 1
+            continue
+
+        # Handle XEN_CPUFEATURE( lines
         res = feat_regex.match(l)
 
         if res is None:
@@ -94,6 +118,15 @@ def parse_definitions(state):
     if len(state.names) == 0:
         raise Fail("No features found")
 
+    if state.nr_entries == 0:
+        raise Fail("No featureset word info found")
+
+    max_val = max(state.names.keys())
+    if (max_val >> 5) >= state.nr_entries:
+        max_name = state.names[max_val]
+        raise Fail("Feature %s (%d*32+%d) exceeds FEATURESET_NR_ENTRIES (%d)"
+                   % (max_name, max_val >> 5, max_val & 31, state.nr_entries))
+
 def featureset_to_uint32s(fs, nr):
     """ Represent a featureset as a list of C-compatible uint32_t's """
 
@@ -122,9 +155,6 @@ def format_uint32s(state, featureset, indent):
 
 def crunch_numbers(state):
 
-    # Size of bitmaps
-    state.nr_entries = nr_entries = (max(state.names.keys()) >> 5) + 1
-
     # Features common between 1d and e1d.
     common_1d = (FPU, VME, DE, PSE, TSC, MSR, PAE, MCE, CX8, APIC,
                  MTRR, PGE, MCA, CMOV, PAT, PSE36, MMX, FXSR)
@@ -328,7 +358,7 @@ def crunch_numbers(state):
     state.nr_deep_deps = len(state.deep_deps.keys())
 
     # Calculate the bitfield name declarations
-    for word in range(nr_entries):
+    for word in range(state.nr_entries):
 
         names = []
         for bit in range(32):
--
generated by git-patchbot for /home/xen/git/xen.git#master