Fix the underlying problem that was causing read(0) to be called: sometimes the

bitcode writer would generate abbrev records saying that the abbrev should be
filled with fixed zero-bit bitfields (this happens in the .bc writer when 
the number of types used in a module is exactly one, since log2(1) == 0).

In this case, just handle it as a literal zero.  We can't "just fix" the writer
without breaking compatibility with existing bc files, so have the abbrev reader
do the substitution.

Strengthen the assert in read to reject reads of zero bits so we catch such 
crimes in the future, and remove the special case designed to handle this.




git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174801 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Chris Lattner 2013-02-09 07:07:29 +00:00
parent e001f27e5e
commit b24f5b7c08
2 changed files with 16 additions and 5 deletions

View File

@ -343,8 +343,8 @@ public:
uint32_t Read(unsigned NumBits) {
assert(NumBits <= 32 && "Cannot return more than 32 bits!");
if (NumBits == 0) return 0;
assert(NumBits && NumBits <= 32 &&
"Cannot return zero or more than 32 bits!");
// If the field is fully contained by CurWord, return it quickly.
if (BitsInCurWord >= NumBits) {

View File

@ -288,9 +288,20 @@ void BitstreamCursor::ReadAbbrevRecord() {
}
BitCodeAbbrevOp::Encoding E = (BitCodeAbbrevOp::Encoding)Read(3);
if (BitCodeAbbrevOp::hasEncodingData(E))
Abbv->Add(BitCodeAbbrevOp(E, ReadVBR64(5)));
else
if (BitCodeAbbrevOp::hasEncodingData(E)) {
unsigned Data = ReadVBR64(5);
// As a special case, handle fixed(0) (i.e., a fixed field with zero bits)
// and vbr(0) as a literal zero. This is decoded the same way, and avoids
// a slow path in Read() to have to handle reading zero bits.
if ((E == BitCodeAbbrevOp::Fixed || E == BitCodeAbbrevOp::VBR) &&
Data == 0) {
Abbv->Add(BitCodeAbbrevOp(0));
continue;
}
Abbv->Add(BitCodeAbbrevOp(E, Data));
} else
Abbv->Add(BitCodeAbbrevOp(E));
}
CurAbbrevs.push_back(Abbv);