I encountered this in the 5.1.54 release branch

Maybe this is a gcc version thing. I use 4.4 in this case. MySQL added -Werror when --with-debug is used. That broke the official build (the problem is in the extra subdir). It also broke changes from Percona

My configure line:
./configure --enable-thread-safe-client --with-plugins=partition,csv,blackhole,myisam,heap,innodb_plugin --without-plugin-innobase --with-fast-mutexes --with-extra-charsets=all --with-debug C_EXTRA_FLAGS="-fno-omit-frame-pointer -fno-strict-aliasing -DHAVE_purify -Wall" CFLAGS="-O0 -g" CXXFLAGS="-g -O0"

cc1: warnings being treated as errors
In file included from ./include/buf0buf.h:1571,
                 from ./include/page0page.h:33,
                 from ./include/page0cur.h:32,
                 from ./include/btr0btr.h:33,
                 from btr/btr0btr.c:26:
./include/buf0buf.ic: In function ‘buf_page_release’:
./include/buf0buf.ic:1078: error: unused parameter ‘mtr’
cc1: warnings being treated as errors
In file included from ./include/buf0buf.h:1571,
                 from ./include/page0page.h:33,
                 from ./include/page0cur.h:32,
                 from ./include/btr0cur.h:31,
                 from btr/btr0cur.c:44:
./include/buf0buf.ic: In function ‘buf_page_release’:
./include/buf0buf.ic:1078: error: unused parameter ‘mtr’
make[2]: *** [libinnobase_a-btr0btr.o] Error 1
make[2]: *** Waiting for unfinished jobs....
make[2]: *** [libinnobase_a-btr0cur.o] Error 1

and another error, I haven't seen this before

cc1: warnings being treated as errors
buf/buf0buddy.c: In function ‘buf_buddy_relocate’:
buf/buf0buddy.c:493: error: ISO C90 forbids mixed declarations and code

the problem is that code was added before declarations

        /* We look inside the allocated objects returned by
        buf_buddy_alloc() and assume that anything of
        PAGE_ZIP_MIN_SIZE or larger is a compressed page that contains
        a valid space_id and page_no in the page header.  Should the
        fields be invalid, we will be unable to relocate the block.
        We also assume that anything that fits sizeof(buf_page_t)
        actually is a properly initialized buf_page_t object. */

        if (size >= PAGE_ZIP_MIN_SIZE) {
                /* This is a compressed page. */
                mutex_t*        mutex;

                if (!have_page_hash_mutex) {
                        mutex_exit(&zip_free_mutex);
                        mutex_enter(&LRU_list_mutex);
                        rw_lock_x_lock(&page_hash_latch);
                }

                /* The src block may be split into smaller blocks,
                some of which may be free.  Thus, the
                mach_read_from_4() calls below may attempt to read
                from free memory.  The memory is "owned" by the buddy
                allocator (and it has been allocated from the buffer
                pool), so there is nothing wrong about this.  The
                mach_read_from_4() calls here will only trigger bogus
                Valgrind memcheck warnings in UNIV_DEBUG_VALGRIND builds. */
                ulint           space   = mach_read_from_4(
                        (const byte*) src + FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID);
                ulint           page_no = mach_read_from_4(
                        (const byte*) src + FIL_PAGE_OFFSET);

After that I gave up and removed Werror from Makefile