These errors can occur if we are built on a system with support for
madvise(MADV_NOFORK) but then we are run on a system whose kernel
does not support that flag.
If the error is something that we don't tolerate at all, we now log
it before crashing.
Fixes bug 31570. I am calling this a bugfix on 0.4.1.1-alpha, where
we actually started using the map_anon code.
--- /dev/null
+ o Major bugfixes (crash, android):
+ - Tolerate systems (including some Android installations) where madvise
+ and MADV_DONTDUMP are available at build-time, but not at run time.
+ Previously, these systems would notice a failed syscall and abort.
+ Fixes bug 31570; bugfix on 0.4.1.1-alpha.
#include <windows.h>
#endif
+#include <string.h>
+#include <errno.h>
+
/**
* Macro to get the high bytes of a size_t, if there are high bytes.
* Windows needs this; other operating systems define a size_t that does
nodump_mem(void *mem, size_t sz)
{
#if defined(MADV_DONTDUMP)
- return madvise(mem, sz, MADV_DONTDUMP);
+ int rv = madvise(mem, sz, MADV_DONTDUMP);
+ if (rv == 0) {
+ return 0;
+ } else if (errno == ENOSYS || errno == EINVAL) {
+ return 0; // syscall not supported, or flag not supported.
+ } else {
+ tor_log_err_sigsafe("Unexpected error from madvise: ",
+ strerror(errno),
+ NULL);
+ return -1;
+ }
#else
(void) mem;
(void) sz;