linux-stable.git/arch/mips/kernel, branch v5.4.76

MIPS: add missing MSACSR and upper MSA initialization

2020-09-09T17:12:28+00:00

[ Upstream commit bb06748207cfb1502d11b90325eba7f8c44c9f02 ]

In cc97ab235f3f ("MIPS: Simplify FP context initialization), init_fp_ctx
just initialize the fp/msa context, and own_fp_inatomic just restore
FCSR and 64bit FP regs from it, but miss MSACSR and upper MSA regs for
MSA, so MSACSR and MSA upper regs's value from previous task on current
cpu can leak into current task and cause unpredictable behavior when MSA
context not initialized.

Fixes: cc97ab235f3f ("MIPS: Simplify FP context initialization")
Signed-off-by: Huang Pei 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin

MIPS: BMIPS: Also call bmips_cpu_setup() for secondary cores

2020-09-09T17:12:24+00:00

[ Upstream commit e14f633b66902615cf7faa5d032b45ab8b6fb158 ]

The initialization done by bmips_cpu_setup() typically affects both
threads of a given core, on 7435 which supports 2 cores and 2 threads,
logical CPU number 2 and 3 would not run this initialization.

Fixes: 738a3f79027b ("MIPS: BMIPS: Add early CPU initialization code")
Signed-off-by: Florian Fainelli 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin

MIPS: Fix unable to reserve memory for Crash kernel

2020-08-26T08:40:55+00:00

[ Upstream commit b1ce9716f3b5ed3b49badf1f003b9e34b7ead0f9 ]

Use 0 as the align parameter in memblock_find_in_range() is
incorrect when we reserve memory for Crash kernel.

The environment as follows:
[    0.000000] MIPS: machine is loongson,loongson64c-4core-rs780e
...
[    1.951016]     crashkernel=64M@128M

The warning as follows:
[    0.000000] Invalid memory region reserved for crash kernel

And the iomem as follows:
00200000-0effffff : System RAM
  04000000-0484009f : Kernel code
  048400a0-04ad7fff : Kernel data
  04b40000-05c4c6bf : Kernel bss
1a000000-1bffffff : pci@1a000000
...

The align parameter may be finally used by round_down() or round_up().
Like the following call tree:

mips-next: mm/memblock.c

memblock_find_in_range
└── memblock_find_in_range_node
    ├── __memblock_find_range_bottom_up
    │   └── round_up
    └── __memblock_find_range_top_down
        └── round_down
\#define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1)
\#define round_down(x, y) ((x) & ~__round_mask(x, y))
\#define __round_mask(x, y) ((__typeof__(x))((y)-1))

The round_down(or round_up)'s second parameter must be a power of 2.
If the second parameter is 0, it both will return 0.

Use 1 as the parameter to fix the bug and the iomem as follows:
00200000-0effffff : System RAM
  04000000-0484009f : Kernel code
  048400a0-04ad7fff : Kernel data
  04b40000-05c4c6bf : Kernel bss
  08000000-0bffffff : Crash kernel
1a000000-1bffffff : pci@1a000000
...

Signed-off-by: Jinyang He 
Reviewed-by: Jiaxun Yang 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin

MIPS: CPU#0 is not hotpluggable

2020-08-21T11:05:26+00:00

commit 9cce844abf07b683cff5f0273977d5f8d0af94c7 upstream.

Now CPU#0 is not hotpluggable on MIPS, so prevent to create /sys/devices
/system/cpu/cpu0/online which confuses some user-space tools.

Cc: stable@vger.kernel.org
Signed-off-by: Huacai Chen 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Greg Kroah-Hartman

MIPS: Add missing EHB in mtc0 -> mfc0 sequence for DSPen

2020-07-09T07:37:56+00:00

commit fcec538ef8cca0ad0b84432235dccd9059c8e6f8 upstream.

This resolves the hazard between the mtc0 in the change_c0_status() and
the mfc0 in configure_exception_vector(). Without resolving this hazard
configure_exception_vector() could read an old value and would restore
this old value again. This would revert the changes change_c0_status()
did. I checked this by printing out the read_c0_status() at the end of
per_cpu_trap_init() and the ST0_MX is not set without this patch.

The hazard is documented in the MIPS Architecture Reference Manual Vol.
III: MIPS32/microMIPS32 Privileged Resource Architecture (MD00088), rev
6.03 table 8.1 which includes:

   Producer | Consumer | Hazard
  ----------|----------|----------------------------
   mtc0     | mfc0     | any coprocessor 0 register

I saw this hazard on an Atheros AR9344 rev 2 SoC with a MIPS 74Kc CPU.
There the change_c0_status() function would activate the DSPen by
setting ST0_MX in the c0_status register. This was reverted and then the
system got a DSP exception when the DSP registers were saved in
save_dsp() in the first process switch. The crash looks like this:

[    0.089999] Mount-cache hash table entries: 1024 (order: 0, 4096 bytes, linear)
[    0.097796] Mountpoint-cache hash table entries: 1024 (order: 0, 4096 bytes, linear)
[    0.107070] Kernel panic - not syncing: Unexpected DSP exception
[    0.113470] Rebooting in 1 seconds..

We saw this problem in OpenWrt only on the MIPS 74Kc based Atheros SoCs,
not on the 24Kc based SoCs. We only saw it with kernel 5.4 not with
kernel 4.19, in addition we had to use GCC 8.4 or 9.X, with GCC 8.3 it
did not happen.

In the kernel I bisected this problem to commit 9012d011660e ("compiler:
allow all arches to enable CONFIG_OPTIMIZE_INLINING"), but when this was
reverted it also happened after commit 172dcd935c34b ("MIPS: Always
allocate exception vector for MIPSr2+").

Commit 0b24cae4d535 ("MIPS: Add missing EHB in mtc0 -> mfc0 sequence.")
does similar changes to a different file. I am not sure if there are
more places affected by this problem.

Signed-off-by: Hauke Mehrtens 
Cc: 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Greg Kroah-Hartman

MIPS: Fix IRQ tracing when call handle_fpe() and handle_msa_fpe()

2020-06-22T07:31:09+00:00

[ Upstream commit 31e1b3efa802f97a17628dde280006c4cee4ce5e ]

Register "a1" is unsaved in this function,
 when CONFIG_TRACE_IRQFLAGS is enabled,
 the TRACE_IRQS_OFF macro will call trace_hardirqs_off(),
 and this may change register "a1".
 The changed register "a1" as argument will be send
 to do_fpe() and do_msa_fpe().

Signed-off-by: YuanJunQing 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin

mips: Add udelay lpj numbers adjustment

2020-06-22T07:31:07+00:00

[ Upstream commit ed26aacfb5f71eecb20a51c4467da440cb719d66 ]

Loops-per-jiffies is a special number which represents a number of
noop-loop cycles per CPU-scheduler quantum - jiffies. As you
understand aside from CPU-specific implementation it depends on
the CPU frequency. So when a platform has the CPU frequency fixed,
we have no problem and the current udelay interface will work
just fine. But as soon as CPU-freq driver is enabled and the cores
frequency changes, we'll end up with distorted udelay's. In order
to fix this we have to accordinly adjust the per-CPU udelay_val
(the same as the global loops_per_jiffy) number. This can be done
in the CPU-freq transition event handler. We subscribe to that event
in the MIPS arch time-inititalization method.

Co-developed-by: Alexey Malahov 
Signed-off-by: Alexey Malahov 
Signed-off-by: Serge Semin 
Reviewed-by: Jiaxun Yang 
Cc: Thomas Bogendoerfer 
Cc: Paul Burton 
Cc: Ralf Baechle 
Cc: Arnd Bergmann 
Cc: Rob Herring 
Cc: devicetree@vger.kernel.org
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin

mips: cm: Fix an invalid error code of INTVN_*_ERR

2020-06-22T07:31:03+00:00

[ Upstream commit 8a0efb8b101665a843205eab3d67ab09cb2d9a8d ]

Commit 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache
errors") adds cm2_causes[] array with map of error type ID and
pointers to the short description string. There is a mistake in
the table, since according to MIPS32 manual CM2_ERROR_TYPE = {17,18}
correspond to INTVN_WR_ERR and INTVN_RD_ERR, while the table
claims they have {0x17,0x18} codes. This is obviously hex-dec
copy-paste bug. Moreover codes {0x18 - 0x1a} indicate L2 ECC errors.

Fixes: 3885c2b463f6 ("MIPS: CM: Add support for reporting CM cache errors")
Signed-off-by: Serge Semin 
Cc: Alexey Malahov 
Cc: Thomas Bogendoerfer 
Cc: Paul Burton 
Cc: Ralf Baechle 
Cc: Arnd Bergmann 
Cc: Rob Herring 
Cc: linux-pm@vger.kernel.org
Cc: devicetree@vger.kernel.org
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin

MIPS: Truncate link address into 32bit for 32bit kernel

2020-06-22T07:31:03+00:00

[ Upstream commit ff487d41036035376e47972c7c522490b839ab37 ]

LLD failed to link vmlinux with 64bit load address for 32bit ELF
while bfd will strip 64bit address into 32bit silently.
To fix LLD build, we should truncate load address provided by platform
into 32bit for 32bit kernel.

Signed-off-by: Jiaxun Yang 
Link: https://github.com/ClangBuiltLinux/linux/issues/786
Link: https://sourceware.org/bugzilla/show_bug.cgi?id=25784
Reviewed-by: Fangrui Song 
Reviewed-by: Kees Cook 
Tested-by: Nathan Chancellor 
Cc: Maciej W. Rozycki 
Tested-by: Nick Desaulniers 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin

MIPS: Make sparse_init() using top-down allocation

2020-06-22T07:30:59+00:00

[ Upstream commit 269b3a9ac538c4ae87f84be640b9fa89914a2489 ]

In the current code, if CONFIG_SWIOTLB is set, when failed to get IO TLB
memory from the low pages by plat_swiotlb_setup(), it may lead to the boot
process failed with kernel panic.

(1) On the Loongson and SiByte platform
arch/mips/loongson64/dma.c
arch/mips/sibyte/common/dma.c
void __init plat_swiotlb_setup(void)
{
	swiotlb_init(1);
}

kernel/dma/swiotlb.c
void  __init
swiotlb_init(int verbose)
{
...
	vstart = memblock_alloc_low(PAGE_ALIGN(bytes), PAGE_SIZE);
	if (vstart && !swiotlb_init_with_tbl(vstart, io_tlb_nslabs, verbose))
		return;
...
	pr_warn("Cannot allocate buffer");
	no_iotlb_memory = true;
}

phys_addr_t swiotlb_tbl_map_single()
{
...
	if (no_iotlb_memory)
		panic("Can not allocate SWIOTLB buffer earlier ...");
...
}

(2) On the Cavium OCTEON platform
arch/mips/cavium-octeon/dma-octeon.c
void __init plat_swiotlb_setup(void)
{
...
	octeon_swiotlb = memblock_alloc_low(swiotlbsize, PAGE_SIZE);
	if (!octeon_swiotlb)
		panic("%s: Failed to allocate %zu bytes align=%lx\n",
		      __func__, swiotlbsize, PAGE_SIZE);
...
}

Because IO_TLB_DEFAULT_SIZE is 64M, if the rest size of low memory is less
than 64M when call plat_swiotlb_setup(), we can easily reproduce the panic
case.

In order to reduce the possibility of kernel panic when failed to get IO
TLB memory under CONFIG_SWIOTLB, it is better to allocate low memory as
small as possible before plat_swiotlb_setup(), so make sparse_init() using
top-down allocation.

Reported-by: Juxin Gao 
Co-developed-by: Juxin Gao 
Signed-off-by: Juxin Gao 
Signed-off-by: Tiezhu Yang 
Signed-off-by: Thomas Bogendoerfer 
Signed-off-by: Sasha Levin