Package: libqb
Version: 0.17.0-2
Severity: important
Tags: patch
Usertags: hppa
libqb fails to build on the hppa architecture, because the built-in testcases
fail as can be seen here:
http://buildd.debian-ports.org/status/fetch.php?pkg=libqb&arch=hppa&ver=0.17.0-2&stamp=1409458262
I did analyzed why they fail, and the reason is that on hppa we have somewhat
more complicated requirements (e.g. alignments) which needs to be followed in
order to mmap shared pages between processes. It's different than what can be
done compared to ia64 and sparc.
The attached patch fixes libqb on the hppa architecture and with it all
testcases finish successful.
It would be great if you could apply this patch (which should not affects other
platforms) to the next upload.
By the way, I fixed a small typo in configure.ac too where arm platforms prints
"ia64"...
Thanks,
Helge
diff -up ./configure.ac.org ./configure.ac
--- ./configure.ac.org 2014-08-31 11:13:34.327159670 +0000
+++ ./configure.ac 2014-09-01 19:22:39.499494241 +0000
@@ -253,10 +253,14 @@ case $host_cpu in
nongcc_memory_barrier_needed=yes
;;
arm*)
- AC_MSG_RESULT([ia64])
+ AC_MSG_RESULT([arm])
AC_DEFINE_UNQUOTED([QB_ARCH_ARM], [1], [arm])
arch_force_shmlba=yes
;;
+ hppa*)
+ AC_MSG_RESULT([hppa])
+ AC_DEFINE_UNQUOTED([QB_ARCH_HPPA], [1], [hppa])
+ ;;
mips*)
AC_MSG_RESULT([ia64])
AC_DEFINE_UNQUOTED([QB_ARCH_MIPS], [1], [mips])
diff -up ./lib/ringbuffer.c.org ./lib/ringbuffer.c
--- ./lib/ringbuffer.c.org 2014-02-20 02:25:14.000000000 +0000
+++ ./lib/ringbuffer.c 2014-09-01 19:22:26.575463547 +0000
@@ -138,7 +138,9 @@ qb_rb_open_2(const char *name, size_t si
void *shm_addr;
long page_size = sysconf(_SC_PAGESIZE);
-#ifdef QB_FORCE_SHM_ALIGN
+#ifdef QB_ARCH_HPPA
+ page_size = QB_MAX(page_size, 0x00400000); /* align to page colour */
+#elif defined(QB_FORCE_SHM_ALIGN)
page_size = QB_MAX(page_size, 16 * 1024);
#endif /* QB_FORCE_SHM_ALIGN */
/* The user of this api expects the 'size' parameter passed into this function
diff -up ./lib/unix.c.org ./lib/unix.c
--- ./lib/unix.c.org 2014-08-31 11:56:35.269419505 +0000
+++ ./lib/unix.c 2014-09-01 19:26:15.696007925 +0000
@@ -170,6 +170,19 @@ qb_sys_circular_mmap(int32_t fd, void **
flags |= MAP_PRIVATE;
#endif /* QB_FORCE_SHM_ALIGN */
+#if defined(QB_ARCH_HPPA)
+ /* map twice the size we want to make sure we have already mapped
+ the second memory location behind it too. Otherwise the Linux
+ kernel may map it in the upper memory so that we can't map
+ the second part afterwards since it will conflict. */
+ addr = mmap(NULL, 2*bytes, PROT_READ | PROT_WRITE,
+ MAP_SHARED, fd, 0);
+
+ if (addr == MAP_FAILED)
+ return -errno;
+
+ addr_orig = addr;
+#else
addr_orig = mmap(NULL, bytes << 1, PROT_NONE, flags, -1, 0);
if (addr_orig == MAP_FAILED) {
@@ -178,6 +191,7 @@ qb_sys_circular_mmap(int32_t fd, void **
addr = mmap(addr_orig, bytes, PROT_READ | PROT_WRITE,
MAP_FIXED | MAP_SHARED, fd, 0);
+#endif
if (addr != addr_orig) {
res = -errno;
