From 9e7a1062327621f551bb897f680da92c2bb9209a Mon Sep 17 00:00:00 2001
From: YAMAMOTO Takashi <yamamoto@midokura.com>
Date: Thu, 19 Jan 2023 12:49:02 +0900
Subject: [PATCH] dlmalloc: require __heap_end

This commit effectively drops the support of older wasm-ld. (LLVM <15.0.7)

We have two relevant use cases:

* `memory.grow` use outside of malloc
  (eg. used by polyfill preview1 binaries)

* `--init-memory` to somehow preallocate heap
  (eg. avoid dynamic allocations, especially on small environments)

While https://github.com/WebAssembly/wasi-libc/pull/377
fixed the former, it broke the latter if you are using
an older LLVM, which doesn't provide the `__heap_end` symbol,
to link your module.

As we couldn't come up with a solution which satisfies all parties,
this commit simply makes it require new enough LLVM which provides
`__heap_end`. After all, a link-time failure is more friendly to users
than failing later in a subtle way.
---
 dlmalloc/src/malloc.c                         | 25 ++++++-------------
 .../wasm32-wasi-threads/undefined-symbols.txt |  1 +
 expected/wasm32-wasi/undefined-symbols.txt    |  1 +
 3 files changed, 10 insertions(+), 17 deletions(-)

diff --git a/dlmalloc/src/malloc.c b/dlmalloc/src/malloc.c
index 6e7a15c03..c6afd938c 100644
--- a/dlmalloc/src/malloc.c
+++ b/dlmalloc/src/malloc.c
@@ -5215,7 +5215,7 @@ static void internal_inspect_all(mstate m,
 
 /* Symbol marking the end of data, bss and explicit stack, provided by wasm-ld. */
 extern char __heap_base;
-extern char __heap_end __attribute__((__weak__));
+extern char __heap_end;
 
 /* Initialize the initial state of dlmalloc to be able to use free memory between __heap_base and initial. */
 static void try_init_allocator(void) {
@@ -5227,23 +5227,14 @@ static void try_init_allocator(void) {
 
   char *base = &__heap_base;
   // Try to use the linker pseudo-symbol `__heap_end` for the initial size of
-  // the heap, but if that's not defined due to LLVM being too old perhaps then
-  // round up `base` to the nearest `PAGESIZE`. The initial size of linear
-  // memory will be at least the heap base to this page boundary, and it's then
-  // assumed that the initial linear memory image was truncated at that point.
-  // While this reflects the default behavior of `wasm-ld` it is also possible
-  // for users to craft larger linear memories by passing options to extend
-  // beyond this threshold. In this situation the memory will not be used for
-  // dlmalloc.
-  //
-  // Note that `sbrk(0)`, or in dlmalloc-ese `CALL_MORECORE(0)`, is specifically
-  // not used here. That captures the current size of the heap but is only
-  // correct if the we're the first to try to grow the heap. If the heap has
-  // grown elsewhere, such as a different allocator in place, then this would
-  // incorrectly claim such memroy as our own.
+  // the heap.
   char *end = &__heap_end;
-  if (end == NULL)
-    end = (char*) page_align((size_t) base);
+  if (end == NULL) {
+    // This can happen when 1. you are using an old wasm-ld which doesn't
+    // provide `__heap_end` and 2. something (other libraries or maybe
+    // your app?) provide a weak reference to `__heap_end`.
+    return;
+  }
   size_t initial_heap_size = end - base;
 
   /* Check that initial heap is long enough to serve a minimal allocation request. */
diff --git a/expected/wasm32-wasi-threads/undefined-symbols.txt b/expected/wasm32-wasi-threads/undefined-symbols.txt
index a68515194..c04b5f703 100644
--- a/expected/wasm32-wasi-threads/undefined-symbols.txt
+++ b/expected/wasm32-wasi-threads/undefined-symbols.txt
@@ -13,6 +13,7 @@ __getf2
 __global_base
 __gttf2
 __heap_base
+__heap_end
 __imported_wasi_snapshot_preview1_args_get
 __imported_wasi_snapshot_preview1_args_sizes_get
 __imported_wasi_snapshot_preview1_clock_res_get
diff --git a/expected/wasm32-wasi/undefined-symbols.txt b/expected/wasm32-wasi/undefined-symbols.txt
index b9cdb3432..6d3b2b7d6 100644
--- a/expected/wasm32-wasi/undefined-symbols.txt
+++ b/expected/wasm32-wasi/undefined-symbols.txt
@@ -11,6 +11,7 @@ __floatunsitf
 __getf2
 __gttf2
 __heap_base
+__heap_end
 __imported_wasi_snapshot_preview1_args_get
 __imported_wasi_snapshot_preview1_args_sizes_get
 __imported_wasi_snapshot_preview1_clock_res_get