1
0
Fork 0
mirror of https://github.com/rubyjs/mini_racer synced 2023-03-27 23:21:28 -04:00

rb_heap_snapshot: fix memory leak from fdopen

The result of fdopen(3) must be matched with fclose(3) to
avoid memory leaks.  However, fclose(3) has the side-effect
of closing the underlying FD behind Ruby's back (which would
corrupt Ruby's internals and cause problems for the remainder
of the process lifetime).

While one could dup(fptr->fd) to fdopen+fclose the resulting fd,
it would cause problems on FD-constrained systems.  Since
GetChunkSize() is available and returns a reasonable value
(64K), the buffering of fwrite(3) is unnecessary and we can
safely use write(2) without performance loss.

So use unbuffered write(2) and perhaps prepare us for better
error reporting in a future change.

I've verified WriteAsciiChunk is indeed receiving the requested
64K (or close to it) in nearly all cases, so excessive syscalls
should not be a problem.
This commit is contained in:
Sam Saffron 2021-12-08 16:17:36 +11:00
parent 2725152c9d
commit 0752264620
No known key found for this signature in database
GPG key ID: B9606168D2FFD9F5

View file

@ -11,6 +11,7 @@
#include <mutex> #include <mutex>
#include <atomic> #include <atomic>
#include <math.h> #include <math.h>
#include <errno.h>
using namespace v8; using namespace v8;
@ -1599,7 +1600,9 @@ rb_heap_stats(VALUE self) {
// https://github.com/bnoordhuis/node-heapdump/blob/master/src/heapdump.cc // https://github.com/bnoordhuis/node-heapdump/blob/master/src/heapdump.cc
class FileOutputStream : public OutputStream { class FileOutputStream : public OutputStream {
public: public:
FileOutputStream(FILE* stream) : stream_(stream) {} int err;
FileOutputStream(int fd) : fd(fd) { err = 0; }
virtual int GetChunkSize() { virtual int GetChunkSize() {
return 65536; return 65536;
@ -1608,17 +1611,27 @@ class FileOutputStream : public OutputStream {
virtual void EndOfStream() {} virtual void EndOfStream() {}
virtual WriteResult WriteAsciiChunk(char* data, int size) { virtual WriteResult WriteAsciiChunk(char* data, int size) {
const size_t len = static_cast<size_t>(size); size_t len = static_cast<size_t>(size);
size_t off = 0;
while (off < len && !feof(stream_) && !ferror(stream_)) while (len) {
off += fwrite(data + off, 1, len - off, stream_); ssize_t w = write(fd, data, len);
return off == len ? kContinue : kAbort; if (w > 0) {
data += w;
len -= w;
} else if (w < 0) {
err = errno;
return kAbort;
} else { /* w == 0, could be out-of-space */
err = -1;
return kAbort;
}
}
return kContinue;
} }
private: private:
FILE* stream_; int fd;
}; };
@ -1631,10 +1644,8 @@ rb_heap_snapshot(VALUE self, VALUE file) {
if (!fptr) return Qfalse; if (!fptr) return Qfalse;
FILE* fp; // prepare for unbuffered write(2) below:
fp = fdopen(fptr->fd, "w"); rb_funcall(file, rb_intern("flush"), 0);
if (fp == NULL) return Qfalse;
ContextInfo* context_info; ContextInfo* context_info;
TypedData_Get_Struct(self, ContextInfo, &context_type, context_info); TypedData_Get_Struct(self, ContextInfo, &context_type, context_info);
@ -1651,13 +1662,14 @@ rb_heap_snapshot(VALUE self, VALUE file) {
const HeapSnapshot* const snap = heap_profiler->TakeHeapSnapshot(); const HeapSnapshot* const snap = heap_profiler->TakeHeapSnapshot();
FileOutputStream stream(fp); FileOutputStream stream(fptr->fd);
snap->Serialize(&stream, HeapSnapshot::kJSON); snap->Serialize(&stream, HeapSnapshot::kJSON);
fflush(fp);
const_cast<HeapSnapshot*>(snap)->Delete(); const_cast<HeapSnapshot*>(snap)->Delete();
/* TODO: perhaps rb_sys_fail here */
if (stream.err) return Qfalse;
return Qtrue; return Qtrue;
} }