Bypass potential libc's sysconf wrappers for sysconf(_SC_PAGESIZE) call

[alekseyshl]

Summary

sysconf(_SC_PAGESIZE) is called very early, during sanitizer init and
any instrumented code (a wrapper/interceptor will likely be instrumented)
calling back to sanitizer before init is done will most surely crash.

[kubamracek]

So is this to avoid potential future crashes when sysconf gets an interceptor? Can we instead call REAL(sysconf)?

[alekseyshl]

No, it's intended to bypass interceptors defined in other libraries. REAL(sysconf) bypasses our own interceptor and REAL(...) functions are not ready yet at the moment we call sysconf(_SC_PAGESIZE) first time anyway.

[kubamracek]

If that something that sanitizers need to protect against, then it sounds like we should do a direct syscall instead. Dlsym is also a complicated thing and it can call other possibly-intercepted functions, e.g. malloc or pthread_mutex_lock, IIRC.

Anyway, I'm not opposing this change, I was just curious why we need it.

[eugenis]

sysconf for the page size is not a syscall on most platforms.
On linux it reads auxv, which could also be done through getauxval() on glibc and bionic, and, it appears, freebsd. But in bionic it is broken.

[alekseyshl]

• Do not call sysconf() during init at all.

[alekseyshl]

It appears that dlsym() also cannot be used during init stage, this code path calls into instrumented code as well under some sanitizers. PTAL.

[eugenis]

LGTM

test/sanitizer_common/TestCases/Linux/sysconf_interceptor_bypass_test.cc

18

Does the test fail w/o this change?
AFAIK CHECK-NOT should go both before and after the CHECK: Passed line.

[alekseyshl]

• Add forgotten CHECK-NOT.

lib/sanitizer_common/sanitizer_linux.cc

 extern char **environ;  // provided by crt1
 #endif  // SANITIZER_FREEBSD
 
 #if !SANITIZER_ANDROID
 #include <sys/signal.h>
 #endif
 
 #if SANITIZER_LINUX
+#include <sys/auxv.h>
 // <linux/time.h>
 struct kernel_timeval {
   long tv_sec;
   long tv_usec;
 };
 
 // <linux/futex.h> is broken on some linux distributions.
 const int FUTEX_WAIT = 0;

 }
 
 uptr GetPageSize() {
 // Android post-M sysconf(_SC_PAGESIZE) crashes if called from .preinit_array.
 #if SANITIZER_ANDROID
   return 4096;
 #elif SANITIZER_LINUX && (defined(__x86_64__) || defined(__i386__))
   return EXEC_PAGESIZE;
+#elif SANITIZER_LINUX
+  return getauxval(AT_PAGESZ);
 #else
   return sysconf(_SC_PAGESIZE);  // EXEC_PAGESIZE may not be trustworthy.
 #endif
 }
 
 uptr ReadBinaryName(/*out*/char *buf, uptr buf_len) {
 #if SANITIZER_FREEBSD
   const int Mib[] = { CTL_KERN, KERN_PROC, KERN_PROC_PATHNAME, -1 };

test/sanitizer_common/TestCases/Linux/sysconf_interceptor_bypass_test.cc

+// RUN: %clangxx -O2 %s -o %t && %run %t 2>&1 | FileCheck %s
+
+#include <stdio.h>
+
+extern "C" long sysconf(int name) {
+  fprintf(stderr, "sysconf wrapper called\n");
+  return 0;
+}
+
+int main() {
+  // All we need to check is that the sysconf() interceptor defined above was
+  // not called. Should it get called, it will crash right there, any
+  // instrumented code executed before sanitizer init is finished will crash
+  // accessing non-initialized sanitizer internals. Even if it will not crash
+  // in some configuration, it should never be called anyway.
+  fprintf(stderr, "Passed\n");
+  // CHECK: Passed
+  // CHECK-NOT: sysconf wrapper called
+  return 0;
+}