Revert fwalk/sfp locking to fix concurrent reads

The locking can fail in a couple of ways:

 - A concurrent fread from an unbuffered or line-buffered file flushes
   the output of other line-buffered files, and if _fwalk locks every
   file, then the fread blocks until other file reads have completed.

 - __sfp can initialize a file lock while _fwalk is locking/unlocking it.

For now, revert to the behavior Bionic had in previous releases. This
commit reverts the file locking parts of commit
468efc80da2504f4ae7de8b5e137426d44dda9d7.

Bug: http://b/131251441
Bug: http://b/130189834
Test: bionic unit tests
Change-Id: I9e20b9cd8ccd14e7962f7308e174f08af72b56c6

1 files changed, 22 insertions, 0 deletions

diff --git a/tests/stdio_test.cpp b/tests/stdio_test.cpp
index 65a942c37..c237d6d32 100644
--- a/tests/stdio_test.cpp
+++ b/tests/stdio_test.cpp
@@ -29,6 +29,7 @@
 #include <locale.h>
 
 #include <string>
+#include <thread>
 #include <vector>
 
 #include <android-base/file.h>
@@ -2577,3 +2578,24 @@ TEST(STDIO_TEST, dev_std_files) {
   ASSERT_LT(0, length);
   ASSERT_EQ("/proc/self/fd/2", std::string(path, length));
 }
+
+TEST(STDIO_TEST, fread_with_locked_file) {
+  // Reading an unbuffered/line-buffered file from one thread shouldn't block on
+  // files locked on other threads, even if it flushes some line-buffered files.
+  FILE* fp1 = fopen("/dev/zero", "r");
+  ASSERT_TRUE(fp1 != nullptr);
+  flockfile(fp1);
+
+  std::thread([] {
+    for (int mode : { _IONBF, _IOLBF }) {
+      FILE* fp2 = fopen("/dev/zero", "r");
+      ASSERT_TRUE(fp2 != nullptr);
+      setvbuf(fp2, nullptr, mode, 0);
+      ASSERT_EQ('\0', fgetc(fp2));
+      fclose(fp2);
+    }
+  }).join();
+
+  funlockfile(fp1);
+  fclose(fp1);
+}