Skip to contentSkip to conversation

s390 patches for the 3.15 merge window

GIT PULL Martin Schwidefsky 2 participants 6 messages Mar 31, 2014 – Apr 2, 2014

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

Hi Linus,

please pull from the 'for-linus' branch of

git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux.git for-linus

to receive the following updates:
There are two memory management related changes, the CMMA support for KVM to avoid swap-in of freed pages and the split page table lock for the PMD level. These two come with common code changes in mm/.

A fix for the long standing theoretical TLB flush problem, this one comes with a common code change in kernel/sched/.

Another set of changes is Heikos uaccess work, included is the initial set of patches with more to come.

And fixes and cleanups as usual.

Dominik Dingel (3):
s390/mm: fixing comment so that parameter name match s390/mm: remove unnecessary parameter from gmap_do_ipte_notify s390/mm: remove unecessary parameter from pgste_ipte_notify

Gerald Schaefer (1):
s390/appldata_os: fix cpu array size calculation

Heiko Carstens (13):
s390/uaccess: normalize order of parameters of indirect uaccess function calls s390/uaccess: get rid of indirect function calls
s390/uaccess: consistent types
s390/setup: get rid of MACHINE_HAS_MVCOS machine flag s390/uaccess: remove dead kernel parameter 'user_mode=' s390/uaccess: introduce 'uaccesspt' kernel parameter s390/bitops: fix comment
s390/uaccess: remove copy_from_user_real()
s390/checksum: remove memset() within csum_partial_copy_from_user() s390/compat: remove compat exec domain
s390/perf: make print_debug_cf() static
s390: update defconfigs
s390/smp: limit number of cpus in possible cpu mask

Josh Triplett (1):
s390: select CONFIG_TTY for use of tty in unconditional keyboard driver

Konstantin Weitz (2):
mm: add support for discard of unused ptes s390/kvm: support collaborative memory management

Martin Schwidefsky (8):
sched/mm: call finish_arch_post_lock_switch in idle_task_exit and use_mm s390/mm,tlb: race of lazy TLB flush vs. recreation of TLB entries s390/kvm: set guest page states to stable on re-ipl
mm: mask bits from pmd in pmd_lockptr/pmd_huge_pte
s390/mm: enable split page table lock for PMD level s390/airq: add support for irq ranges
s390/ptrace: add support for PTRACE_SINGLEBLOCK s390/con3270: optionally disable auto update

Michael Holzheu (2):
s390/sclp_early: Return correct HSA block count also for zero
hypfs: Add clarification for "weight_min" attribute

Philipp Hachtmann (1):
s390/topology: Remove call to update_cpu_masks()

Sebastian Ott (5):
s390/cio: fix driver callback initialization for ccw consoles s390/cio: reorder initialization of ccw consoles s390/cio: fix irq stats for early interrupts on ccw consoles
s390: improve debug feature usage
s390: add some drivers/subsystems to the MAINTAINERS file

 MAINTAINERS                             |   27 ++++-
 arch/s390/Kconfig                       |    5 +
 arch/s390/appldata/appldata_os.c        |    2 +-
 arch/s390/configs/default_defconfig     |   46 +++++++--
 arch/s390/configs/gcov_defconfig        |   33 ++++--
 arch/s390/configs/performance_defconfig |   33 ++++--
 arch/s390/configs/zfcpdump_defconfig    |    1 -
 arch/s390/defconfig                     |   10 ++
 arch/s390/hypfs/hypfs_vm.c              |    9 +-
 arch/s390/include/asm/airq.h            |   14 ++-
 arch/s390/include/asm/bitops.h          |    8 +-
 arch/s390/include/asm/ccwdev.h          |    4 +-
 arch/s390/include/asm/checksum.h        |   11 +-
 arch/s390/include/asm/futex.h           |   13 +--
 arch/s390/include/asm/kvm_host.h        |    5 +-
 arch/s390/include/asm/mmu_context.h     |   39 ++++++-
 arch/s390/include/asm/pgalloc.h         |   18 +++-
 arch/s390/include/asm/pgtable.h         |  105 +++++++++++++------
 arch/s390/include/asm/ptrace.h          |    1 +
 arch/s390/include/asm/sclp.h            |    1 +
 arch/s390/include/asm/setup.h           |    3 -
 arch/s390/include/asm/thread_info.h     |    3 +
 arch/s390/include/asm/uaccess.h         |  171 +++++++++++++------------------
 arch/s390/include/uapi/asm/ptrace.h     |    6 ++
 arch/s390/kernel/Makefile               |    5 +-
 arch/s390/kernel/compat_exec_domain.c   |   29 ------
 arch/s390/kernel/early.c                |    2 -
 arch/s390/kernel/entry.S                |    9 +-
 arch/s390/kernel/entry64.S              |    9 +-
 arch/s390/kernel/perf_event.c           |    2 +-
 arch/s390/kernel/ptrace.c               |   13 ++-
 arch/s390/kernel/setup.c                |   17 ---
 arch/s390/kernel/smp.c                  |    8 +-
 arch/s390/kernel/topology.c             |    1 -
 arch/s390/kvm/diag.c                    |    3 +
 arch/s390/kvm/kvm-s390.c                |   25 +++++
 arch/s390/kvm/kvm-s390.h                |    2 +
 arch/s390/kvm/priv.c                    |   41 ++++++++
 arch/s390/lib/Makefile                  |    3 +-
 arch/s390/lib/find.c                    |    2 +-
 arch/s390/lib/uaccess.h                 |    8 +-
 arch/s390/lib/uaccess_mvcos.c           |  116 ++++++++++++++++-----
 arch/s390/lib/uaccess_pt.c              |   69 +++++--------
 arch/s390/mm/maccess.c                  |   28 +----
 arch/s390/mm/pgtable.c                  |  158 +++++++++++++++++++++++++++-
 arch/s390/pci/pci_debug.c               |    2 +-
 drivers/s390/char/con3215.c             |    8 +-
 drivers/s390/char/con3270.c             |   13 ++-
 drivers/s390/char/raw3270.c             |   26 ++++-
 drivers/s390/char/raw3270.h             |    3 +-
 drivers/s390/char/sclp_early.c          |   31 +++++-
 drivers/s390/cio/airq.c                 |   66 +++++++-----
 drivers/s390/cio/chsc_sch.c             |    3 +-
 drivers/s390/cio/cio.c                  |    4 +-
 drivers/s390/cio/device.c               |   52 ++++++----
 drivers/s390/net/qeth_core_main.c       |    4 +-
 include/asm-generic/pgtable.h           |   13 +++
 include/linux/mm.h                      |   10 +-
 kernel/sched/core.c                     |    4 +-
 mm/mmu_context.c                        |    3 +
 mm/rmap.c                               |   10 ++
 61 files changed, 943 insertions(+), 427 deletions(-)
 delete mode 100644 arch/s390/kernel/compat_exec_domain.c
#1 lore raw
Quoted 4 lines · Martin Schwidefsky

There are two memory management related changes, the CMMA support for KVM to avoid swap-in of freed pages and the split page table lock for the PMD level. These two come with common code changes in mm/.

Ugh. I pulled it, but things like this makes me want to dig my eyes out with a spoon:

+#ifdef finish_arch_post_lock_switch
+ finish_arch_post_lock_switch();
+#endif

when I think the proper thing to do would have been to move the #ifndef that creates an empty finish_arch_post_lock_switch() from kernel/sched/sched.h to some common file, or possibly even just duplicate it.

I detest #ifdef's in the middle of code. Yes, we do have them, but we should try to avoid adding more of them.

Maybe we could have a <linux/mmu_context.h> that includes the <asm/mmu_context.h> and then does that "let's add the dummy function for architectures that don't need it"?

Added Ingo to the cc, since this ends up intersecting with the scheduler code that now does the wrapper on its own.

Linus

#2 lore raw
Quoted 22 lines · On Mon, 31 Mar 2014 14:45:32 -0700 Linus Torvalds…
Quoted 4 lines · Martin Schwidefsky

There are two memory management related changes, the CMMA support for KVM to avoid swap-in of freed pages and the split page table lock for the PMD level. These two come with common code changes in mm/.

Ugh. I pulled it, but things like this makes me want to dig my eyes out with a spoon:

+#ifdef finish_arch_post_lock_switch
+ finish_arch_post_lock_switch();
+#endif

when I think the proper thing to do would have been to move the #ifndef that creates an empty finish_arch_post_lock_switch() from kernel/sched/sched.h to some common file, or possibly even just duplicate it.

I detest #ifdef's in the middle of code. Yes, we do have them, but we should try to avoid adding more of them.

Quoted 6 lines

Maybe we could have a <linux/mmu_context.h> that includes the <asm/mmu_context.h> and then does that "let's add the dummy function for architectures that don't need it"?

Added Ingo to the cc, since this ends up intersecting with the scheduler code that now does the wrapper on its own.

Yes, I feel the pain. I actually tried to move the finish_arch_post_lock_switch to linux/mmu_context.h but it turned out that this does not work. It broke the corgi_defconfig, this is what I got from the friendly kbuild test robot:

-- snip

tree:   git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux.git features
head:   05bb2956f28ff01a102bf16702ba7bcb116ecf72
commit: 8f6d84ba48d8d4377c61477803c8ff2cd4b007b7 [10/14] sched/mm: call finish_arch_post_lock_switch in idle_task_exit and use_mm
config: make ARCH=arm corgi_defconfig

All error/warnings:

In file included from include/linux/mmu_context.h:5:0,
from drivers/usb/gadget/inode.c:27:
arch/arm/include/asm/mmu_context.h: In function 'finish_arch_post_lock_switch':

Quoted 1 line
Quoted 1 line

arch/arm/include/asm/mmu_context.h:82:3: error: implicit declaration of function 'preempt_enable_no_resched' [-Werror=implicit-function-declaration]

cc1: some warnings being treated as errors

vim +/preempt_enable_no_resched +82 arch/arm/include/asm/mmu_context.h

bdae73cd arch/arm/include/asm/mmu_context.h Catalin Marinas 2013-07-23  76  		 */
bdae73cd arch/arm/include/asm/mmu_context.h Catalin Marinas 2013-07-23  77  		preempt_disable();
bdae73cd arch/arm/include/asm/mmu_context.h Catalin Marinas 2013-07-23  78  		if (mm->context.switch_pending) {
bdae73cd arch/arm/include/asm/mmu_context.h Catalin Marinas 2013-07-23  79  			mm->context.switch_pending = 0;
bdae73cd arch/arm/include/asm/mmu_context.h Catalin Marinas 2013-07-23  80  			cpu_switch_mm(mm->pgd, mm);
bdae73cd arch/arm/include/asm/mmu_context.h Catalin Marinas 2013-07-23  81  		}
bdae73cd arch/arm/include/asm/mmu_context.h Catalin Marinas 2013-07-23 @82  		preempt_enable_no_resched();
b9d4d42a arch/arm/include/asm/mmu_context.h Catalin Marinas 2011-11-28  83  	}
b9d4d42a arch/arm/include/asm/mmu_context.h Catalin Marinas 2011-11-28  84  }

^1da177e include/asm-arm/mmu_context.h Linus Torvalds 2005-04-16 85

:::::: The code at line 82 was first introduced by commit
:::::: bdae73cd374e28db544fdd9b77de689a36e3c129 ARM: 7790/1: Fix deferred mm switch on VIVT processors

:::::: TO: Catalin Marinas <[email protected]> :::::: CC: Russell King <[email protected]>

-- snip

I gave up after a few hours trying to solve the header dependencies, the two trouble makes are alpha with the task_thread_info call in mmu_context.h and arm with the preempt_enable_no_resched call. Welcome to include file hell.

Signature · 4 lines
blue skies,
   Martin.

"Reality continues to ruin my life." - Calvin.
#3 lore raw
Quoted 4 lines · Martin Schwidefsky

I gave up after a few hours trying to solve the header dependencies, the two trouble makes are alpha with the task_thread_info call in mmu_context.h and arm with the preempt_enable_no_resched call. Welcome to include file hell.

I don't understand. The only file <asm/mmu_context.h> should need is <linux/mm.h>. Maybe <kernel/sched.h>.

Look at mm/mmu_context.c, for example. It includes <asm/mmu_context.h> without having included very many header files at all. So we *know* that that small set is sufficient.

And there aren't very many users of <linux/mmu_context.h> and none of them are other header files, so there's no point in trying to religiously avoid including one or two header files from it.

So I'm thinking something like this should work. It's untested, and maybe some architecture really needs some other header file, but it looks quite likely to work. No?

Linus

---
 include/linux/mmu_context.h | 8 +++++++-
 kernel/sched/core.c         | 2 +-
 kernel/sched/sched.h        | 3 ---
 mm/mmu_context.c            | 4 ----
 4 files changed, 8 insertions(+), 9 deletions(-)
include/linux/mmu_context.h +7 -1
index 70fffeba7495..843bbc2abec0 100644
@@ -1,7 +1,13 @@
 #ifndef _LINUX_MMU_CONTEXT_H
 #define _LINUX_MMU_CONTEXT_H
 
-struct mm_struct;
+#include <linux/mm.h>
+#include <linux/sched.h>
+#include <asm/mmu_context.h>
+
+#ifndef finish_arch_post_lock_switch
+#  define finish_arch_post_lock_switch() do { } while (0)
+#endif
 
 void use_mm(struct mm_struct *mm);
 void unuse_mm(struct mm_struct *mm);
kernel/sched/core.c +1 -1
index a47902c687ae..20fc1bd8aec4 100644
@@ -32,7 +32,7 @@
 #include <linux/init.h>
 #include <linux/uaccess.h>
 #include <linux/highmem.h>
-#include <asm/mmu_context.h>
+#include <linux/mmu_context.h>
 #include <linux/interrupt.h>
 #include <linux/capability.h>
 #include <linux/completion.h>
kernel/sched/sched.h +0 -3
index f2de7a175620..5eafd5e8c236 100644
@@ -960,9 +960,6 @@ static inline int task_running(struct rq *rq, struct task_struct *p)
 #ifndef finish_arch_switch
 # define finish_arch_switch(prev)	do { } while (0)
 #endif
-#ifndef finish_arch_post_lock_switch
-# define finish_arch_post_lock_switch()	do { } while (0)
-#endif
 
 #ifndef __ARCH_WANT_UNLOCKED_CTXSW
 static inline void prepare_lock_switch(struct rq *rq, struct task_struct *next)
mm/mmu_context.c +0 -4
index f802c2d216a7..56ecbbdbf364 100644
@@ -8,8 +8,6 @@
 #include <linux/export.h>
 #include <linux/sched.h>
 
-#include <asm/mmu_context.h>
-
 /*
  * use_mm
  *	Makes the calling kernel thread take on the specified
@@ -31,9 +29,7 @@ void use_mm(struct mm_struct *mm)
 	tsk->mm = mm;
 	switch_mm(active_mm, mm, tsk);
 	task_unlock(tsk);
-#ifdef finish_arch_post_lock_switch
 	finish_arch_post_lock_switch();
-#endif
 
 	if (active_mm != mm)
 		mmdrop(active_mm);
#4 lore raw
Quoted 15 lines · On Tue, 1 Apr 2014 08:40:41 -0700 (PDT) Linus Torvalds…
Quoted 4 lines · Martin Schwidefsky

I gave up after a few hours trying to solve the header dependencies, the two trouble makes are alpha with the task_thread_info call in mmu_context.h and arm with the preempt_enable_no_resched call. Welcome to include file hell.

I don't understand. The only file <asm/mmu_context.h> should need is <linux/mm.h>. Maybe <kernel/sched.h>.

Look at mm/mmu_context.c, for example. It includes <asm/mmu_context.h> without having included very many header files at all. So we *know* that that small set is sufficient.

Quoted 7 lines

And there aren't very many users of <linux/mmu_context.h> and none of them are other header files, so there's no point in trying to religiously avoid including one or two header files from it.

So I'm thinking something like this should work. It's untested, and maybe some architecture really needs some other header file, but it looks quite likely to work. No?

I had a patch quite similar to yours and I ended up with the error report from the kbuild robot. Just tried yours with the corgi_defconfig and got this:

In file included from include/linux/mmu_context.h:6,
from drivers/usb/gadget/inode.c:27:
/home4/mschwide/linux/arch/arm/include/asm/mmu_context.h: In function 'finish_arch_post_lock_switch':
/home4/mschwide/linux/arch/arm/include/asm/mmu_context.h:82: error: implicit declaration of function 'preempt_enable_no_resched'

Part of the problem is this lovely piece from include/linux/preempt.h:

#ifdef MODULE
/*
* Modules have no business playing preemption tricks. */
#undef sched_preempt_enable_no_resched
#undef preempt_enable_no_resched
#undef preempt_enable_no_resched_notrace #undef preempt_check_resched
#endif

Signature · 4 lines
blue skies,
   Martin.

"Reality continues to ruin my life." - Calvin.
#5 lore raw
Quoted 2 lines · Martin Schwidefsky

Part of the problem is this lovely piece from include/linux/preempt.h:

Oh, the pain. That's annoying. Oh well. I guess we'll live with the ugliness for now.

Linus