linux.git/arch, branch v5.4.148

s390/bpf: Fix 64-bit subtraction of the -0x80000000 constant

2021-09-22T10:26:46+00:00

commit 6e61dc9da0b7a0d91d57c2e20b5ea4fd2d4e7e53 upstream.

The JIT uses agfi for subtracting constants, but -(-0x80000000) cannot
be represented as a 32-bit signed binary integer. Fix by using algfi in
this particular case.

Reported-by: Johan Almbladh 
Fixes: 054623105728 ("s390/bpf: Add s390x eBPF JIT compiler backend")
Reviewed-by: Heiko Carstens 
Signed-off-by: Ilya Leoshkevich 
Signed-off-by: Vasily Gorbik 
Signed-off-by: Greg Kroah-Hartman

s390/bpf: Fix optimizing out zero-extensions

2021-09-22T10:26:46+00:00

commit db7bee653859ef7179be933e7d1384644f795f26 upstream.

Currently the JIT completely removes things like `reg32 += 0`,
however, the BPF_ALU semantics requires the target register to be
zero-extended in such cases.

Fix by optimizing out only the arithmetic operation, but not the
subsequent zero-extension.

Reported-by: Johan Almbladh 
Fixes: 054623105728 ("s390/bpf: Add s390x eBPF JIT compiler backend")
Reviewed-by: Heiko Carstens 
Signed-off-by: Ilya Leoshkevich 
Signed-off-by: Vasily Gorbik 
Signed-off-by: Greg Kroah-Hartman

ARC: export clear_user_page() for modules

2021-09-22T10:26:45+00:00

[ Upstream commit 6b5ff0405e4190f23780362ea324b250bc495683 ]

0day bot reports a build error:
  ERROR: modpost: "clear_user_page" [drivers/media/v4l2-core/videobuf-dma-sg.ko] undefined!
so export it in arch/arc/ to fix the build error.

In most ARCHes, clear_user_page() is a macro. OTOH, in a few
ARCHes it is a function and needs to be exported.
PowerPC exported it in 2004. It looks like nds32 and nios2
still need to have it exported.

Fixes: 4102b53392d63 ("ARC: [mm] Aliasing VIPT dcache support 2/4")
Signed-off-by: Randy Dunlap 
Reported-by: kernel test robot 
Cc: Guenter Roeck 
Cc: linux-snps-arc@lists.infradead.org
Signed-off-by: Vineet Gupta 
Signed-off-by: Sasha Levin

KVM: PPC: Book3S HV: Tolerate treclaim. in fake-suspend mode changing registers

2021-09-22T10:26:42+00:00

commit 267cdfa21385d78c794768233678756e32b39ead upstream.

POWER9 DD2.2 and 2.3 hardware implements a "fake-suspend" mode where
certain TM instructions executed in HV=0 mode cause softpatch interrupts
so the hypervisor can emulate them and prevent problematic processor
conditions. In this fake-suspend mode, the treclaim. instruction does
not modify registers.

Unfortunately the rfscv instruction executed by the guest do not
generate softpatch interrupts, which can cause the hypervisor to lose
track of the fake-suspend mode, and it can execute this treclaim. while
not in fake-suspend mode. This modifies GPRs and crashes the hypervisor.

It's not trivial to disable scv in the guest with HFSCR now, because
they assume a POWER9 has scv available. So this fix saves and restores
checkpointed registers across the treclaim.

Fixes: 7854f7545bff ("KVM: PPC: Book3S: Rework TM save/restore code and make it C-callable")
Signed-off-by: Nicholas Piggin 
Signed-off-by: Michael Ellerman 
Link: https://lore.kernel.org/r/20210908101718.118522-2-npiggin@gmail.com
Signed-off-by: Greg Kroah-Hartman

x86/mm: Fix kern_addr_valid() to cope with existing but not present entries

2021-09-22T10:26:40+00:00

commit 34b1999da935a33be6239226bfa6cd4f704c5c88 upstream.

Jiri Olsa reported a fault when running:

  # cat /proc/kallsyms | grep ksys_read
  ffffffff8136d580 T ksys_read
  # objdump -d --start-address=0xffffffff8136d580 --stop-address=0xffffffff8136d590 /proc/kcore

  /proc/kcore:     file format elf64-x86-64

  Segmentation fault

  general protection fault, probably for non-canonical address 0xf887ffcbff000: 0000 [#1] SMP PTI
  CPU: 12 PID: 1079 Comm: objdump Not tainted 5.14.0-rc5qemu+ #508
  Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.14.0-4.fc34 04/01/2014
  RIP: 0010:kern_addr_valid
  Call Trace:
   read_kcore
   ? rcu_read_lock_sched_held
   ? rcu_read_lock_sched_held
   ? rcu_read_lock_sched_held
   ? trace_hardirqs_on
   ? rcu_read_lock_sched_held
   ? lock_acquire
   ? lock_acquire
   ? rcu_read_lock_sched_held
   ? lock_acquire
   ? rcu_read_lock_sched_held
   ? rcu_read_lock_sched_held
   ? rcu_read_lock_sched_held
   ? lock_release
   ? _raw_spin_unlock
   ? __handle_mm_fault
   ? rcu_read_lock_sched_held
   ? lock_acquire
   ? rcu_read_lock_sched_held
   ? lock_release
   proc_reg_read
   ? vfs_read
   vfs_read
   ksys_read
   do_syscall_64
   entry_SYSCALL_64_after_hwframe

The fault happens because kern_addr_valid() dereferences existent but not
present PMD in the high kernel mappings.

Such PMDs are created when free_kernel_image_pages() frees regions larger
than 2Mb. In this case, a part of the freed memory is mapped with PMDs and
the set_memory_np_noalias() -> ... -> __change_page_attr() sequence will
mark the PMD as not present rather than wipe it completely.

Have kern_addr_valid() check whether higher level page table entries are
present before trying to dereference them to fix this issue and to avoid
similar issues in the future.

Stable backporting note:
------------------------

Note that the stable marking is for all active stable branches because
there could be cases where pagetable entries exist but are not valid -
see 9a14aefc1d28 ("x86: cpa, fix lookup_address"), for example. So make
sure to be on the safe side here and use pXY_present() accessors rather
than pXY_none() which could #GP when accessing pages in the direct map.

Also see:

  c40a56a7818c ("x86/mm/init: Remove freed kernel image areas from alias mapping")

for more info.

Reported-by: Jiri Olsa 
Signed-off-by: Mike Rapoport 
Signed-off-by: Borislav Petkov 
Reviewed-by: David Hildenbrand 
Acked-by: Dave Hansen 
Tested-by: Jiri Olsa 
Cc: 	# 4.4+
Link: https://lkml.kernel.org/r/20210819132717.19358-1-rppt@kernel.org
Signed-off-by: Greg Kroah-Hartman

arm64/sve: Use correct size when reinitialising SVE state

2021-09-22T10:26:39+00:00

commit e35ac9d0b56e9efefaeeb84b635ea26c2839ea86 upstream.

When we need a buffer for SVE register state we call sve_alloc() to make
sure that one is there. In order to avoid repeated allocations and frees
we keep the buffer around unless we change vector length and just memset()
it to ensure a clean register state. The function that deals with this
takes the task to operate on as an argument, however in the case where we
do a memset() we initialise using the SVE state size for the current task
rather than the task passed as an argument.

This is only an issue in the case where we are setting the register state
for a task via ptrace and the task being configured has a different vector
length to the task tracing it. In the case where the buffer is larger in
the traced process we will leak old state from the traced process to
itself, in the case where the buffer is smaller in the traced process we
will overflow the buffer and corrupt memory.

Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")
Cc:  # 4.15.x
Signed-off-by: Mark Brown 
Link: https://lore.kernel.org/r/20210909165356.10675-1-broonie@kernel.org
Signed-off-by: Catalin Marinas 
Signed-off-by: Greg Kroah-Hartman

xen: reset legacy rtc flag for PV domU

2021-09-22T10:26:39+00:00

commit f68aa100d815b5b4467fd1c3abbe3b99d65fd028 upstream.

A Xen PV guest doesn't have a legacy RTC device, so reset the legacy
RTC flag. Otherwise the following WARN splat will occur at boot:

[    1.333404] WARNING: CPU: 1 PID: 1 at /home/gross/linux/head/drivers/rtc/rtc-mc146818-lib.c:25 mc146818_get_time+0x1be/0x210
[    1.333404] Modules linked in:
[    1.333404] CPU: 1 PID: 1 Comm: swapper/0 Tainted: G        W         5.14.0-rc7-default+ #282
[    1.333404] RIP: e030:mc146818_get_time+0x1be/0x210
[    1.333404] Code: c0 64 01 c5 83 fd 45 89 6b 14 7f 06 83 c5 64 89 6b 14 41 83 ec 01 b8 02 00 00 00 44 89 63 10 5b 5d 41 5c 41 5d 41 5e 41 5f c3 <0f> 0b 48 c7 c7 30 0e ef 82 4c 89 e6 e8 71 2a 24 00 48 c7 c0 ff ff
[    1.333404] RSP: e02b:ffffc90040093df8 EFLAGS: 00010002
[    1.333404] RAX: 00000000000000ff RBX: ffffc90040093e34 RCX: 0000000000000000
[    1.333404] RDX: 0000000000000001 RSI: 0000000000000000 RDI: 000000000000000d
[    1.333404] RBP: ffffffff82ef0e30 R08: ffff888005013e60 R09: 0000000000000000
[    1.333404] R10: ffffffff82373e9b R11: 0000000000033080 R12: 0000000000000200
[    1.333404] R13: 0000000000000000 R14: 0000000000000002 R15: ffffffff82cdc6d4
[    1.333404] FS:  0000000000000000(0000) GS:ffff88807d440000(0000) knlGS:0000000000000000
[    1.333404] CS:  10000e030 DS: 0000 ES: 0000 CR0: 0000000080050033
[    1.333404] CR2: 0000000000000000 CR3: 000000000260a000 CR4: 0000000000050660
[    1.333404] Call Trace:
[    1.333404]  ? wakeup_sources_sysfs_init+0x30/0x30
[    1.333404]  ? rdinit_setup+0x2b/0x2b
[    1.333404]  early_resume_init+0x23/0xa4
[    1.333404]  ? cn_proc_init+0x36/0x36
[    1.333404]  do_one_initcall+0x3e/0x200
[    1.333404]  kernel_init_freeable+0x232/0x28e
[    1.333404]  ? rest_init+0xd0/0xd0
[    1.333404]  kernel_init+0x16/0x120
[    1.333404]  ret_from_fork+0x1f/0x30

Cc: 
Fixes: 8d152e7a5c7537 ("x86/rtc: Replace paravirt rtc check with platform legacy quirk")
Signed-off-by: Juergen Gross 
Reviewed-by: Boris Ostrovsky 
Link: https://lore.kernel.org/r/20210903084937.19392-3-jgross@suse.com
Signed-off-by: Juergen Gross 
Signed-off-by: Greg Kroah-Hartman

s390/pv: fix the forcing of the swiotlb

2021-09-22T10:26:37+00:00

commit 93ebb6828723b8aef114415c4dc3518342f7dcad upstream.

Since commit 903cd0f315fe ("swiotlb: Use is_swiotlb_force_bounce for
swiotlb data bouncing") if code sets swiotlb_force it needs to do so
before the swiotlb is initialised. Otherwise
io_tlb_default_mem->force_bounce will not get set to true, and devices
that use (the default) swiotlb will not bounce despite switolb_force
having the value of SWIOTLB_FORCE.

Let us restore swiotlb functionality for PV by fulfilling this new
requirement.

This change addresses what turned out to be a fragility in
commit 64e1f0c531d1 ("s390/mm: force swiotlb for protected
virtualization"), which ain't exactly broken in its original context,
but could give us some more headache if people backport the broken
change and forget this fix.

Signed-off-by: Halil Pasic 
Tested-by: Christian Borntraeger 
Reviewed-by: Christian Borntraeger 
Fixes: 903cd0f315fe ("swiotlb: Use is_swiotlb_force_bounce for swiotlb data bouncing")
Fixes: 64e1f0c531d1 ("s390/mm: force swiotlb for protected virtualization")
Cc: stable@vger.kernel.org #5.3+
Signed-off-by: Konrad Rzeszutek Wilk 
Signed-off-by: Greg Kroah-Hartman

parisc: fix crash with signals and alloca

2021-09-22T10:26:37+00:00

commit 030f653078316a9cc9ca6bd1b0234dcf858be35d upstream.

I was debugging some crashes on parisc and I found out that there is a
crash possibility if a function using alloca is interrupted by a signal.
The reason for the crash is that the gcc alloca implementation leaves
garbage in the upper 32 bits of the sp register. This normally doesn't
matter (the upper bits are ignored because the PSW W-bit is clear),
however the signal delivery routine in the kernel uses full 64 bits of sp
and it fails with -EFAULT if the upper 32 bits are not zero.

I created this program that demonstrates the problem:

#include 
#include 
#include 
#include 

static __attribute__((noinline,noclone)) void aa(int *size)
{
	void * volatile p = alloca(-*size);
	while (1) ;
}

static void handler(int sig)
{
	write(1, "signal delivered\n", 17);
	_exit(0);
}

int main(void)
{
	int size = -0x100;
	signal(SIGALRM, handler);
	alarm(1);
	aa(&size);
}

If you compile it with optimizations, it will crash.
The "aa" function has this disassembly:

000106a0 :
   106a0:       08 03 02 41     copy r3,r1
   106a4:       08 1e 02 43     copy sp,r3
   106a8:       6f c1 00 80     stw,ma r1,40(sp)
   106ac:       37 dc 3f c1     ldo -20(sp),ret0
   106b0:       0c 7c 12 90     stw ret0,8(r3)
   106b4:       0f 40 10 9c     ldw 0(r26),ret0		; ret0 = 0x00000000FFFFFF00
   106b8:       97 9c 00 7e     subi 3f,ret0,ret0	; ret0 = 0xFFFFFFFF0000013F
   106bc:       d7 80 1c 1a     depwi 0,31,6,ret0	; ret0 = 0xFFFFFFFF00000100
   106c0:       0b 9e 0a 1e     add,l sp,ret0,sp	;   sp = 0xFFFFFFFFxxxxxxxx
   106c4:       e8 1f 1f f7     b,l,n 106c4 ,r0

This patch fixes the bug by truncating the "usp" variable to 32 bits.

Signed-off-by: Mikulas Patocka 
Cc: stable@vger.kernel.org
Signed-off-by: Helge Deller 
Signed-off-by: Greg Kroah-Hartman

m68knommu: only set CONFIG_ISA_DMA_API for ColdFire sub-arch

2021-09-22T10:26:34+00:00

[ Upstream commit db87db65c1059f3be04506d122f8ec9b2fa3b05e ]

> Hi Arnd,
>
> First bad commit (maybe != root cause):
>
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
> head:   2f73937c9aa561e2082839bc1a8efaac75d6e244
> commit: 47fd22f2b84765a2f7e3f150282497b902624547 [4771/5318] cs89x0: rework driver configuration
> config: m68k-randconfig-c003-20210804 (attached as .config)
> compiler: m68k-linux-gcc (GCC) 10.3.0
> reproduce (this is a W=1 build):
>         wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
>         chmod +x ~/bin/make.cross
>         # https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=47fd22f2b84765a2f7e3f150282497b902624547
>         git remote add linux-next https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git
>         git fetch --no-tags linux-next master
>         git checkout 47fd22f2b84765a2f7e3f150282497b902624547
>         # save the attached .config to linux build tree
>         COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-10.3.0 make.cross ARCH=m68k
>
> If you fix the issue, kindly add following tag as appropriate
> Reported-by: kernel test robot 
>
> All errors (new ones prefixed by >>):
>
>    In file included from include/linux/kernel.h:19,
>                     from include/linux/list.h:9,
>                     from include/linux/module.h:12,
>                     from drivers/net/ethernet/cirrus/cs89x0.c:51:
>    drivers/net/ethernet/cirrus/cs89x0.c: In function 'net_open':
>    drivers/net/ethernet/cirrus/cs89x0.c:897:20: error: implicit declaration of function 'isa_virt_to_bus'; did you mean 'virt_to_bus'? [-Werror=implicit-function-declaration]
>      897 |     (unsigned long)isa_virt_to_bus(lp->dma_buff));
>          |                    ^~~~~~~~~~~~~~~
>    include/linux/printk.h:141:17: note: in definition of macro 'no_printk'
>      141 |   printk(fmt, ##__VA_ARGS__);  \
>          |                 ^~~~~~~~~~~
>    drivers/net/ethernet/cirrus/cs89x0.c:86:3: note: in expansion of macro 'pr_debug'
>       86 |   pr_##level(fmt, ##__VA_ARGS__);   \
>          |   ^~~
>    drivers/net/ethernet/cirrus/cs89x0.c:894:3: note: in expansion of macro 'cs89_dbg'
>      894 |   cs89_dbg(1, debug, "%s: dma %lx %lx\n",
>          |   ^~~~~~~~
> >> drivers/net/ethernet/cirrus/cs89x0.c:914:3: error: implicit declaration of function 'disable_dma'; did you mean 'disable_irq'? [-Werror=implicit-function-declaration]

As far as I can tell, this is a bug with the m68kmmu architecture, not
with my driver:
The CONFIG_ISA_DMA_API option is provided for coldfire, which implements it,
but dragonball also sets the option as a side-effect, without actually
implementing
the interfaces. The patch below should fix it.

Signed-off-by: Arnd Bergmann 
Signed-off-by: Greg Ungerer 
Signed-off-by: Sasha Levin