This patch implements handling of STRICT_KERNEL_RWX with
large TLBs directly in the TLB miss handlers.
To do so, etext and sinittext are aligned on 512kB boundaries
and the miss handlers use 512kB pages instead of 8Mb pages for
addresses close to the boundaries.
It sets RO PP flags for addresses under sinittext.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
At the time being, mmu_mapin_ram() always maps RAM from the beginning.
But some platforms like the WII have to map a second block of RAM.
This patch adds to mmu_mapin_ram() the base address of the block.
At the moment, only base address 0 is supported.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
At the time being, initial MMU setup allows 24 Mbytes
of DATA and 8 Mbytes of code.
Some debug setup like CONFIG_KASAN generate huge
kernels with text size over the 8M limit and data over the
24 Mbytes limit.
Here is an 8xx kernel compiled with CONFIG_KASAN_INLINE for
one of my boards:
[root@po16846vm linux-powerpc]# size -x vmlinux
text data bss dec hex filename
0x111019c 0x41b0d4 0x490de0 26984528 19bc050 vmlinux
This patch maps up to 32 Mbytes code based on _einittext symbol
and allows 32 Mbytes of memory instead of 24.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This patch replaces most #ifdef mess by IS_ENABLED() in 8xx_mmu.c
This has the advantage of allowing syntax verification at compile
time regardless of selected options.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
There is no reason to re-read each time the pointer at
location 0xf0 as it is fixed and known.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Instead of hardcoding the TLB handlers patching, use
the newly created modify_instruction_site() helper.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Today, on the 8xx the TLB handlers do SW tablewalk by doing all
the calculation in ASM, in order to match with the Linux page
table structure.
The 8xx offers hardware assistance which allows significant size
reduction of the TLB handlers, hence also reduces the time spent
in the handlers.
However, using this HW assistance implies some constraints on the
page table structure:
- Regardless of the main page size used (4k or 16k), the
level 1 table (PGD) contains 1024 entries and each PGD entry covers
a 4Mbytes area which is managed by a level 2 table (PTE) containing
also 1024 entries each describing a 4k page.
- 16k pages require 4 identifical entries in the L2 table
- 512k pages PTE have to be spread every 128 bytes in the L2 table
- 8M pages PTE are at the address pointed by the L1 entry and each
8M page require 2 identical entries in the PGD.
This patch modifies the TLB handlers to use HW assistance for 4K PAGES.
Before that patch, the mean time spent in TLB miss handlers is:
- ITLB miss: 80 ticks
- DTLB miss: 62 ticks
After that patch, the mean time spent in TLB miss handlers is:
- ITLB miss: 72 ticks
- DTLB miss: 54 ticks
So the improvement is 10% for ITLB and 13% for DTLB misses
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The 8xx TLB miss routines are patched at startup at several places.
This patch uses the new patch_site functionality in order
to get a better code readability and avoid a label mess when
dumping the code with 'objdump -d'
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This reverts commit 4f94b2c746.
That commit was buggy, as it used rlwinm instead of rlwimi.
Instead of fixing that bug, we revert the previous commit in order to
reduce the dependency between L1 entries and L2 entries
Fixes: 4f94b2c746 ("powerpc/8xx: Use L1 entry APG to handle _PAGE_ACCESSED for CONFIG_SWAP")
Cc: stable@vger.kernel.org
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
_PAGE_PRIVILEGED corresponds to the SH bit which doesn't protect
against user access but only disables ASID verification on kernel
accesses. User access is controlled with _PMD_USER flag.
Name it _PAGE_SH instead of _PAGE_PRIVILEGED
_PAGE_HUGE corresponds to the SPS bit which doesn't really tells
that's it is a huge page but only that it is not a 4k page.
Name it _PAGE_SPS instead of _PAGE_HUGE
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
In order to avoid multiple conversions, handover directly a
pgprot_t to map_kernel_page() as already done for radix.
Do the same for __ioremap_caller() and __ioremap_at().
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On the 8xx, the page size is set in the PMD entry and applies to
all pages of the page table pointed by the said PMD entry.
When an app has some regular pages allocated (e.g. see below) and tries
to mmap() a huge page at a hint address covered by the same PMD entry,
the kernel accepts the hint allthough the 8xx cannot handle different
page sizes in the same PMD entry.
10000000-10001000 r-xp 00000000 00:0f 2597 /root/malloc
10010000-10011000 rwxp 00000000 00:0f 2597 /root/malloc
mmap(0x10080000, 524288, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_ANONYMOUS|0x40000, -1, 0) = 0x10080000
This results the app remaining forever in do_page_fault()/hugetlb_fault()
and when interrupting that app, we get the following warning:
[162980.035629] WARNING: CPU: 0 PID: 2777 at arch/powerpc/mm/hugetlbpage.c:354 hugetlb_free_pgd_range+0xc8/0x1e4
[162980.035699] CPU: 0 PID: 2777 Comm: malloc Tainted: G W 4.14.6 #85
[162980.035744] task: c67e2c00 task.stack: c668e000
[162980.035783] NIP: c000fe18 LR: c00e1eec CTR: c00f90c0
[162980.035830] REGS: c668fc20 TRAP: 0700 Tainted: G W (4.14.6)
[162980.035854] MSR: 00029032 <EE,ME,IR,DR,RI> CR: 24044224 XER: 20000000
[162980.036003]
[162980.036003] GPR00: c00e1eec c668fcd0 c67e2c00 00000010 c6869410 10080000 00000000 77fb4000
[162980.036003] GPR08: ffff0001 0683c001 00000000 ffffff80 44028228 10018a34 00004008 418004fc
[162980.036003] GPR16: c668e000 00040100 c668e000 c06c0000 c668fe78 c668e000 c6835ba0 c668fd48
[162980.036003] GPR24: 00000000 73ffffff 74000000 00000001 77fb4000 100fffff 10100000 10100000
[162980.036743] NIP [c000fe18] hugetlb_free_pgd_range+0xc8/0x1e4
[162980.036839] LR [c00e1eec] free_pgtables+0x12c/0x150
[162980.036861] Call Trace:
[162980.036939] [c668fcd0] [c00f0774] unlink_anon_vmas+0x1c4/0x214 (unreliable)
[162980.037040] [c668fd10] [c00e1eec] free_pgtables+0x12c/0x150
[162980.037118] [c668fd40] [c00eabac] exit_mmap+0xe8/0x1b4
[162980.037210] [c668fda0] [c0019710] mmput.part.9+0x20/0xd8
[162980.037301] [c668fdb0] [c001ecb0] do_exit+0x1f0/0x93c
[162980.037386] [c668fe00] [c001f478] do_group_exit+0x40/0xcc
[162980.037479] [c668fe10] [c002a76c] get_signal+0x47c/0x614
[162980.037570] [c668fe70] [c0007840] do_signal+0x54/0x244
[162980.037654] [c668ff30] [c0007ae8] do_notify_resume+0x34/0x88
[162980.037744] [c668ff40] [c000dae8] do_user_signal+0x74/0xc4
[162980.037781] Instruction dump:
[162980.037821] 7fdff378 81370000 54a3463a 80890020 7d24182e 7c841a14 712a0004 4082ff94
[162980.038014] 2f890000 419e0010 712a0ff0 408200e0 <0fe00000> 54a9000a 7f984840 419d0094
[162980.038216] ---[ end trace c0ceeca8e7a5800a ]---
[162980.038754] BUG: non-zero nr_ptes on freeing mm: 1
[162985.363322] BUG: non-zero nr_ptes on freeing mm: -1
In order to fix this, this patch uses the address space "slices"
implemented for BOOK3S/64 and enhanced to support PPC32 by the
preceding patch.
This patch modifies the context.id on the 8xx to be in the range
[1:16] instead of [0:15] in order to identify context.id == 0 as
not initialised contexts as done on BOOK3S
This patch activates CONFIG_PPC_MM_SLICES when CONFIG_HUGETLB_PAGE is
selected for the 8xx
Alltough we could in theory have as many slices as PMD entries, the
current slices implementation limits the number of low slices to 16.
This limitation is not preventing us to fix the initial issue allthough
it is suboptimal. It will be cured in a subsequent patch.
Fixes: 4b91428699 ("powerpc/8xx: Implement support of hugepages")
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When CONFIG_SWAP is set, the TLB miss handlers have to also take
into account _PAGE_ACCESSED flag. At the moment it is done by
anding _PAGE_ACCESSED into _PAGE_PRESENT using 3 instructions.
This patch uses APG for handling _PAGE_ACCESSED, allowing to
just copy _PAGE_ACCESSED bit into APG field, hence reducing the
action to a single instruction.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
commit ac29c64089 ("powerpc/mm: Replace _PAGE_USER with
_PAGE_PRIVILEGED") introduced _PAGE_PRIVILEGED for BOOK3S/64
This patch generalises _PAGE_PRIVILEGED for all CPUs, allowing
to have either _PAGE_PRIVILEGED or _PAGE_USER or both.
PPC_8xx has a _PAGE_SHARED flag which is set for and only for
all non user pages. Lets rename it _PAGE_PRIVILEGED to remove
confusion as it has nothing to do with Linux shared pages.
On BookE, there's a _PAGE_BAP_SR which has to be set for kernel
pages: defining _PAGE_PRIVILEGED as _PAGE_BAP_SR will make
this generic
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
setup_initial_memory_limit() is only called during init.
mmu_patch_cmp_limit() is only called from 8xx_mmu.c
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
As stated in a comment in head_8xx.S, today we "Always pin the first
8 MB ITLB to prevent ITLB misses while mucking around with SRR0/SRR1
in asm".
This issue has just been cleared by the preceding patch, therefore
we can make this pinning optional (on by default) and independent
of DATA pinning.
This patch also makes pinning of IMMR independent of pinning of DATA.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On the 8xx, the RAM mapped with LTLBs must be seen as block mapped,
just like areas mapped with BATs on standard PPC32.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
These two functions implement the same semantics, so unify their naming so we
can share code that calls them. The longer name is more descriptive so use it.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Acked-by: Balbir Singh <bsingharora@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
CONFIG_PIN_TLB maps IMMR area and the first 24 Mbytes of memory.
In some circunstances it might be more interesting to not map
IMMR but map 32 Mbytes of memory instead.
Therefore we add config option CONFIG_PIN_TLB_IMMR to select if
IMMR shall be pinned or not, hence whether we pin 24 or 32 Mbytes of RAM
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
On recent kernels, with some debug options like for instance
CONFIG_LOCKDEP, the BSS requires more than 8M memory, allthough
the kernel code fits in the first 8M.
Today, it is necessary to activate CONFIG_PIN_TLB to get more than 8M
at startup, allthough pinning TLB is not necessary for that.
We could have inconditionaly mapped 16 or 24M bytes at startup
but some old hardware only have 8M and mapping non-existing RAM
would be an issue due to speculative accesses.
With the preceding patch however, the TLB entries are populated on
demand. By setting up the TLB miss handler to handle up to 24M until
the handler is patched for the entire memory space, it is possible
to allow access up to more memory without mapping non-existing RAM.
It is therefore not needed anymore to map memory data at all
at startup. It will be handled by the TLB miss handler.
One might still want to PIN the IMMR and the first 24M of RAM.
It is now possible to do it in the C memory initialisation
functions. In addition, we now know how much memory we have
when we do it, so we are able to adapt the pining to the
real amount of memory available. So boards with less than 24M
can now also benefit from PIN_TLB.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
Instead of using the first level page table to define mappings for
the linear memory space, we can use direct mapping from the TLB
handling routines. This has several advantages:
* No need to read the tables at each TLB miss
* No issue in 16k pages mode where the 1st level table maps 64 Mbytes
The size of the available linear space is known at system startup.
In order to avoid data access at each TLB miss to know the memory
size, the TLB routine is patched at startup with the proper size
This patch provides a 10%-15% improvment of TLB miss handling for
kernel addresses
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
Once the linear memory space has been mapped with 8Mb pages, as
seen in the related commit, we get 11 millions DTLB missed during
the reference 600s period. 77% of the misses are on user addresses
and 23% are on kernel addresses (1 fourth for linear address space
and 3 fourth for virtual address space)
Traditionaly, each driver manages one computer board which has its
own components with its own memory maps.
But on embedded chips like the MPC8xx, the SOC has all registers
located in the same IO area.
When looking at ioremaps done during startup, we see that
many drivers are re-mapping small parts of the IMMR for their own use
and all those small pieces gets their own 4k page, amplifying the
number of TLB misses: in our system we get 0xff000000 mapped 31 times
and 0xff003000 mapped 9 times.
Even if each part of IMMR was mapped only once with 4k pages, it would
still be several small mappings towards linear area.
This patch maps the IMMR with a single 512k page.
With this patch applied, the number of DTLB misses during the 10 min
period is reduced to 11.8 millions for a duration of 5.8s, which
represents 2% of the non-idle time hence yet another 10% reduction.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
On PPC8xx, flushing instruction cache is performed by writing
in register SPRN_IC_CST. This registers suffers CPU6 ERRATA.
The patch rewrites the fonction in C so that CPU6 ERRATA will
be handled transparently
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
There is no real need to have set_context() in assembly.
Now that we have mtspr() handling CPU6 ERRATA directly, we
can rewrite set_context() in C language for easier maintenance.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
Now we have a 8xx specific .c file for that so put it in there
as other powerpc variants do
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
On a live running system (VoIP gateway for Air Trafic Control), over
a 10 minutes period (with 277s idle), we get 87 millions DTLB misses
and approximatly 35 secondes are spent in DTLB handler.
This represents 5.8% of the overall time and even 10.8% of the
non-idle time.
Among those 87 millions DTLB misses, 15% are on user addresses and
85% are on kernel addresses. And within the kernel addresses, 93%
are on addresses from the linear address space and only 7% are on
addresses from the virtual address space.
MPC8xx has no BATs but it has 8Mb page size. This patch implements
mapping of kernel RAM using 8Mb pages, on the same model as what is
done on the 40x.
In 4k pages mode, each PGD entry maps a 4Mb area: we map every two
entries to the same 8Mb physical page. In each second entry, we add
4Mb to the page physical address to ease life of the FixupDAR
routine. This is just ignored by HW.
In 16k pages mode, each PGD entry maps a 64Mb area: each PGD entry
will point to the first page of the area. The DTLB handler adds
the 3 bits from EPN to map the correct page.
With this patch applied, we now get only 13 millions TLB misses
during the 10 minutes period. The idle time has increased to 313s
and the overall time spent in DTLB miss handler is 6.3s, which
represents 1% of the overall time and 2.2% of non-idle time.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
Christophe Leroy