Don't treat "main" thread specially. This simplifies access to
thread-specific data.
xmlGetGlobalState can now also fail on the former main thread, leading
to an unrecoverable condition if malloc fails.
The globals were never defined in public header files when compiling
with thread support. Now they're only defined in a legacy build.
Move TlsFree to DllMain to make cleanup more robust on Windows.
Obsoletes #1.
Port most changes made to the xmlBuf code in f3807d76, except that
"size" still includes the terminating NULL byte.
Make xmlSetBufferAllocationScheme, xmlBufferAllocScheme and
xmlDefaultBufferSize no-ops.
Deprecate a few functions.
Implement XML_SAVE_NO_INDENT to disable and XML_SAVE_INDENT to enable
indenting regardless of the global xmlIndentTreeOutput.
Implement XML_SAVE_EMPTY to enable empty tags regardless of the global
xmlSaveNoEmptyTags.
See #736.
On Linux, we tried to detect the presence of libpthread to disable
things like locks. This questionable hack doesn't work since glibc 2.34
which merged libpthread into libc.
Introduce xmlCtxtSetErrorHandler allowing to set a structured error for
a parser context. There already was the "serror" SAX handler but this
always receives the parser context as argument.
Start to use xmlRaiseMemoryError.
Remove useless arguments from memory error functions. Rename
xmlErrMemory to xmlCtxtErrMemory.
Remove a few calls to xmlGenericError.
Remove support for runtime entity debugging.
Setting these deprecated globals hasn't had an effect for a long time.
Make them constants. This reduces the size of per-thread storage from
~700 to ~250 bytes.
Linking executables will fail on systems with glibc < 2.34 without
declaring these symbols as weak references.
In commit c19771c1f13de9196f98260d142d8c8672eb5733 these references
were moved to globals.c from threads.c, but the `#pragma weak`
declarations were lost in the process.
Also removing unneeded weak declarations from threads.c.
It seems that thread-local storage destructors are run before pthread
thread-specific data destructors on Darwin, defeating our scheme to use
TSD to clean up TLS.
Here's an example program that reports a use-after-free when compiled
with `-fsanitize=address` on macOS:
#include <pthread.h>
typedef struct {
int v;
} my_struct;
static _Thread_local my_struct tls;
pthread_key_t key;
void dtor(void *tsd) {
my_struct *s = (my_struct *) tsd;
/*
* This will crash ASan, apparently because
* TLS has already been freed.
*/
s->v = 1;
}
void *thread(void *p) {
pthread_setspecific(key, &tls);
return NULL;
}
int main(void) {
pthread_key_create(&key, dtor);
pthread_t handle;
pthread_create(&handle, NULL, thread, NULL);
pthread_join(handle, NULL);
return 0;
}
