2 files changed, 92 insertions, 20 deletions
diff --git a/Documentation/technical/api-merge.txt b/Documentation/technical/api-merge.txt
index a7e050bb7a..9dc1bed768 100644
--- a/Documentation/technical/api-merge.txt
+++ b/Documentation/technical/api-merge.txt
@@ -17,6 +17,40 @@ responsible for a few things.
    path-specific merge drivers (specified in `.gitattributes`)
    into account.
 
+Data structures
+---------------
+
+* `mmbuffer_t`, `mmfile_t`
+
+These store data usable for use by the xdiff backend, for writing and
+for reading, respectively.  See `xdiff/xdiff.h` for the definitions
+and `diff.c` for examples.
+
+* `struct ll_merge_options`
+
+This describes the set of options the calling program wants to affect
+the operation of a low-level (single file) merge.  Some options:
+
+`virtual_ancestor`::
+	Behave as though this were part of a merge between common
+	ancestors in a recursive merge.
+	If a helper program is specified by the
+	`[merge "<driver>"] recursive` configuration, it will
+	be used (see linkgit:gitattributes[5]).
+
+`variant`::
+	Resolve local conflicts automatically in favor
+	of one side or the other (as in 'git merge-file'
+	`--ours`/`--theirs`/`--union`).  Can be `0`,
+	`XDL_MERGE_FAVOR_OURS`, `XDL_MERGE_FAVOR_THEIRS`, or
+	`XDL_MERGE_FAVOR_UNION`.
+
+`renormalize`::
+	Resmudge and clean the "base", "theirs" and "ours" files
+	before merging.  Use this when the merge is likely to have
+	overlapped with a change in smudge/clean or end-of-line
+	normalization rules.
+
 Low-level (single file) merge
 -----------------------------
 
@@ -28,15 +62,24 @@ Low-level (single file) merge
 	`.git/info/attributes` into account.  Returns 0 for a
 	clean merge.
 
-The caller:
+Calling sequence:
 
-1. allocates an mmbuffer_t variable for the result;
-2. allocates and fills variables with the file's original content
-   and two modified versions (using `read_mmfile`, for example);
-3. calls ll_merge();
-4. reads the output from result_buf.ptr and result_buf.size;
-5. releases buffers when finished (free(ancestor.ptr); free(ours.ptr);
-   free(theirs.ptr); free(result_buf.ptr);).
+* Prepare a `struct ll_merge_options` to record options.
+  If you have no special requests, skip this and pass `NULL`
+  as the `opts` parameter to use the default options.
+
+* Allocate an mmbuffer_t variable for the result.
+
+* Allocate and fill variables with the file's original content
+  and two modified versions (using `read_mmfile`, for example).
+
+* Call `ll_merge()`.
+
+* Read the merged content from `result_buf.ptr` and `result_buf.size`.
+
+* Release buffers when finished.  A simple
+  `free(ancestor.ptr); free(ours.ptr); free(theirs.ptr);
+  free(result_buf.ptr);` will do.
 
 If the modifications do not merge cleanly, `ll_merge` will return a
 nonzero value and `result_buf` will generally include a description of
@@ -47,18 +90,6 @@ The `ancestor_label`, `our_label`, and `their_label` parameters are
 used to label the different sides of a conflict if the merge driver
 supports this.
 
-The `flag` parameter is a bitfield:
-
- - The `LL_OPT_VIRTUAL_ANCESTOR` bit indicates whether this is an
-   internal merge to consolidate ancestors for a recursive merge.
-
- - The `LL_OPT_FAVOR_MASK` bits allow local conflicts to be automatically
-   resolved in favor of one side or the other (as in 'git merge-file'
-   `--ours`/`--theirs`/`--union`).
-   They can be populated by `create_ll_flag`, whose argument can be
-   `XDL_MERGE_FAVOR_OURS`, `XDL_MERGE_FAVOR_THEIRS`, or
-   `XDL_MERGE_FAVOR_UNION`.
-
 Everything else
 ---------------
 
diff --git a/Documentation/technical/api-sigchain.txt b/Documentation/technical/api-sigchain.txt
new file mode 100644
index 0000000000..535cdff164
--- /dev/null
+++ b/Documentation/technical/api-sigchain.txt
@@ -0,0 +1,41 @@
+sigchain API
+============
+
+Code often wants to set a signal handler to clean up temporary files or
+other work-in-progress when we die unexpectedly. For multiple pieces of
+code to do this without conflicting, each piece of code must remember
+the old value of the handler and restore it either when:
+
+  1. The work-in-progress is finished, and the handler is no longer
+     necessary. The handler should revert to the original behavior
+     (either another handler, SIG_DFL, or SIG_IGN).
+
+  2. The signal is received. We should then do our cleanup, then chain
+     to the next handler (or die if it is SIG_DFL).
+
+Sigchain is a tiny library for keeping a stack of handlers. Your handler
+and installation code should look something like:
+
+------------------------------------------
+  void clean_foo_on_signal(int sig)
+  {
+	  clean_foo();
+	  sigchain_pop(sig);
+	  raise(sig);
+  }
+
+  void other_func()
+  {
+	  sigchain_push_common(clean_foo_on_signal);
+	  mess_up_foo();
+	  clean_foo();
+  }
+------------------------------------------
+
+Handlers are given the typdef of sigchain_fun. This is the same type
+that is given to signal() or sigaction(). It is perfectly reasonable to
+push SIG_DFL or SIG_IGN onto the stack.
+
+You can sigchain_push and sigchain_pop individual signals. For
+convenience, sigchain_push_common will push the handler onto the stack
+for many common signals.