- 22 May, 2014 1 commit
-
-
Scott Wood authored
The only way Freescale booke chips support mappings larger than 4K is via TLB1. The only way we support (direct) TLB1 entries is via hugetlb, which is not what map_kernel_page() does when given a large page size. Without this, a kernel with CONFIG_SPARSEMEM_VMEMMAP enabled crashes on boot with messages such as: PID hash table entries: 4096 (order: 3, 32768 bytes) Sorting __ex_table... BUG: Bad page state in process swapper pfn:00a2f page:8000040000023a48 count:0 mapcount:0 mapping:0000040000ffce48 index:0x40000ffbe50 page flags: 0x40000ffda40(active|arch_1|private|private_2|head|tail|swapcache|mappedtodisk|reclaim|swapbacked|unevictable|mlocked) page dumped because: PAGE_FLAGS_CHECK_AT_FREE flag(s) set bad because of flags: page flags: 0x311840(active|private|private_2|swapcache|unevictable|mlocked) Modules linked in: CPU: 0 PID: 0 Comm: swapper Not tainted 3.15.0-rc1-00003-g7fa250c #299 Call Trace: [c00000000098ba20] [c000000000008b3c] .show_stack+0x7c/0x1cc (unreliable) [c00000000098baf0] [c00000000060aa50] .dump_stack+0x88/0xb4 [c00000000098bb70] [c0000000000c0468] .bad_page+0x144/0x1a0 [c00000000098bc10] [c0000000000c0628] .free_pages_prepare+0x164/0x17c [c00000000098bcc0] [c0000000000c24cc] .free_hot_cold_page+0x48/0x214 [c00000000098bd60] [c00000000086c318] .free_all_bootmem+0x1fc/0x354 [c00000000098be70] [c00000000085da84] .mem_init+0xac/0xdc [c00000000098bef0] [c0000000008547b0] .start_kernel+0x21c/0x4d4 [c00000000098bf90] [c000000000000448] .start_here_common+0x20/0x58 Signed-off-by:Scott Wood <scottwood@freescale.com>
-
- 09 May, 2014 1 commit
-
-
Scott Wood authored
Several of the error paths from fsl_rio_setup are missing error messages. Worse, fsl_rio_setup initializes several global pointers and does not NULL them out after freeing/unmapping on error. This caused fsl_rio_mcheck_exception() to crash when accessing rio_regs_win which was non-NULL but had been unmapped. Signed-off-by:
-
- 05 May, 2014 1 commit
-
-
Benjamin Herrenschmidt authored
This series adds support for building the powerpc 64-bit LE kernel using the new ABI v2. We already supported running ABI v2 userspace programs but this adds support for building the kernel itself using the new ABI.
-
- 30 Apr, 2014 8 commits
-
-
Liu Ping Fan authored
In case of extending the eaddr in future, use this macro PGTABLE_EADDR_SIZE to ease the maintenance of the code. Signed-off-by:
Liu Ping Fan <pingfank@linux.vnet.ibm.com> Signed-off-by:
Benjamin Herrenschmidt <benh@kernel.crashing.org>
-
Alistair Popple authored
This patch adds an option to enable a work around for an icache bug on 476 that can cause execution of stale instructions when falling through pages (IBM errata #46). It requires a recent version of binutils which supports the --ppc476-workaround option. The work around enables the appropriate linker options and ensures that all module output sections are aligned to 4K page boundaries. The work around is only required when building modules. Signed-off-by:
Alistair Popple <alistair@popple.id.au> Signed-off-by:
-
Alistair Popple authored
The PPC476GTR SoC supports message signalled interrupts (MSI) by writing to special addresses within the High Speed Transfer Assist (HSTA) module. This patch adds support for PCI MSI with a new system device. The DMA window is also updated to allow access to the entire 42-bit address range to allow PCI devices write access to the HSTA module. Signed-off-by:
-
Alistair Popple authored
This patch adds support for the IBM Akebono board. Signed-off-by:
-
Alistair Popple authored
The IBM Akebono code uses the same initialisation functions as the earlier Currituck board. Rather than create a copy of this code for Akebono we will instead integrate support for it into the same file as the Currituck code. This patch just renames the board support file and updates the Makefile. Signed-off-by:
-
Tony Breeds authored
Signed-off-by:
Tony Breeds <tony@bakeyournoodle.com> Signed-off-by:
-
Alexander Graf authored
When we never get around to seeing an HEA ethernet adapter, there's no point in restricting ourselves to 4k IO page size. This speeds up IO maps when CONFIG_IBMEBUS is disabled. [ Updated the test to also lift the restriction on arch 2.07 (Power 8) which cannot have an HEA -- BenH ] Signed-off-by:
Alexander Graf <agraf@suse.de> Signed-off-by:
-
Philippe Bergheaud authored
Unaligned stores take alignment exceptions on POWER7 running in little-endian. This is a dumb little-endian base memcpy that prevents unaligned stores. Once booted the feature fixup code switches over to the VMX copy loops (which are already endian safe). The question is what we do before that switch over. The base 64bit memcpy takes alignment exceptions on POWER7 so we can't use it as is. Fixing the causes of alignment exception would slow it down, because we'd need to ensure all loads and stores are aligned either through rotate tricks or bytewise loads and stores. Either would be bad for all other 64bit platforms. [ I simplified the loop a bit - Anton ] Signed-off-by:
Philippe Bergheaud <felix@linux.vnet.ibm.com> Signed-off-by:
Anton Blanchard <anton@samba.org> Signed-off-by:
-
- 29 Apr, 2014 1 commit
-
-
Benjamin Herrenschmidt authored
Merge Linus tree to get "cpufreq, powernv: Fix build failure on UP" to avoid build breakages in some of my test configs.
-
- 28 Apr, 2014 28 commits
-
-
Michael Neuling authored
If we do a treclaim and we are not in TM suspend mode, it results in a TM bad thing (ie. a 0x700 program check). Similarly if we do a trechkpt and we have an active transaction or TEXASR Failure Summary (FS) is not set, we also take a TM bad thing. This should never happen, but if it does (ie. a kernel bug), the cause is almost impossible to debug as the GPR state is mostly userspace and hence we don't get a call chain. This adds some checks in these cases case a BUG_ON() (in asm) in case we ever hit these cases. It moves the register saving around to preserve r1 till later also. Signed-off-by:
Michael Neuling <mikey@neuling.org> Signed-off-by:
-
Michael Neuling authored
We save r1 to the scratch SPR and restore it from there after the trechkpt so saving r1 to the paca is not needed. Signed-off-by:
-
Gautham R. Shenoy authored
Implement a method named pnv_get_proc_freq(unsigned int cpu) which returns the current clock rate on the 'cpu' in Hz to be reported in /proc/cpuinfo. This method uses the value reported by cpufreq when such a value is sane. Otherwise it falls back to old way of reporting the clockrate, i.e. ppc_proc_freq. Set the ppc_md.get_proc_freq() hook to pnv_get_proc_freq() on the PowerNV platform. Signed-off-by:
Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by:
-
Gautham R. Shenoy authored
Currently, the code in setup-common.c for powerpc assumes that all clock rates are same in a smp system. This value is cached in the variable named ppc_proc_freq and is the value that is reported in /proc/cpuinfo. However on the PowerNV platform, the clock rate is same only across the threads of the same core. Hence the value that is reported in /proc/cpuinfo is incorrect on PowerNV platforms. We need a better way to query and report the correct value of the processor clock in /proc/cpuinfo. The patch achieves this by creating a machdep_call named get_proc_freq() which is expected to returns the frequency in Hz. The code in show_cpuinfo() can invoke this method to display the correct clock rate on platforms that have implemented this method. On the other powerpc platforms it can use the value cached in ppc_proc_freq. Signed-off-by:
-
Vasant Hegde authored
Firmware update on PowerNV platform takes several minutes. During this time one CPU is stuck in FW and the kernel complains about "soft lockups". This patch returns all secondary CPUs to firmware before starting firmware update process. [ Reworked a bit and cleaned up -- BenH ] Signed-off-by:
Vasant Hegde <hegdevasant@linux.vnet.ibm.com> Signed-off-by:
-
Andrew Murray authored
This patch updates the implementation of pci_process_bridge_OF_ranges to use the of_pci_range_parser helpers. Signed-off-by:
Andrew Murray <amurray@embedded-bits.co.uk> Signed-off-by:
-
Stephen Chivers authored
This patch adds support to legacy serial for UARTS with shifted registers. The MVME5100 Single Board Computer is a PowerPC platform that has 16550 style UARTS with register addresses that are 16 bytes apart (shifted by 4). Commit 30925748 "powerpc: Cleanup udbg_16550 and add support for LPC PIO-only UARTs" added support to udbg_16550 for shifted registers by adding a "stride" parameter to the initialisation operations for Programmed IO and Memory Mapped IO. As a consequence it is now possible to use the services of legacy serial to provide early serial console messages for the MVME5100. An added benefit of this is that the serial console will always be "ttyS0" irrespective of whether the computer is fitted with extra PCI 8250 interface boards or not. I have tested this patch using the four PowerPC platforms available to me: MVME5100 - shifted registers, SAM440EP - unshifted registers, MPC8349 - unshifted registers, MVME4100 - unshifted registers. Signed-off-by:
Stephen Chivers <schivers@csc.com> Signed-off-by:
-
Cédric Le Goater authored
The code is only slightly modified : entry points now use the FIXUP_ENDIAN trampoline to switch endian order. The 32bit wrapper is kept for big endian kernels and 64bit is enforced for little endian kernels with a PPC64_BOOT_WRAPPER config option. The linker script is generated using the kernel preprocessor flags to make use of the CONFIG_* definitions and the wrapper script is modified to take into account the new elf64ppc format. Finally, the zImage file is compiled as a position independent executable (-pie) which makes it loadable at any address by the firmware. Signed-off-by:
Cédric Le Goater <clg@fr.ibm.com> Signed-off-by:
-
Cédric Le Goater authored
When entering the boot wrapper in little endian, we will need to fix the endian order using a fixup trampoline like in the kernel. This patch overrides the _zimage_start entry point for this purpose. Signed-off-by:
-
Cédric Le Goater authored
This patch adds support a 64bit wrapper entry point. As in 32bit, the entry point does its own relocation and can be loaded at any address by the firmware. Signed-off-by:
-
Cédric Le Goater authored
This patch defines a 'prom' routine similar to 'enter_prom' in the kernel. The difference is in the MSR which is built before entering prom. Big endian order is enforced as in the kernel but 32bit mode is not. It prepares ground for the next patches which will introduce Little endian order. Signed-off-by:
-
Cédric Le Goater authored
Signed-off-by:
-
Cédric Le Goater authored
It could certainly be improved using Elf macros and byteswapping routines, but the initial version of the code is organised to be a single file program with limited dependencies. yaboot is the same. Please scream if you want a total rewrite. Signed-off-by:
-
Cédric Le Goater authored
These are not the most efficient versions of swab but the wrapper does not do much byte swapping. On a big endian cpu, these routines are a no-op. Signed-off-by:
-
Cédric Le Goater authored
arch/powerpc/boot/oflib.c:211:9: warning: cast to pointer from integer of \ different size [-Wint-to-pointer-cast] return (phandle) of_call_prom("finddevice", 1, 1, name); This is a work around. The definite solution would be to define the phandle typedef as a u32, as in the kernel, but this would break the device tree ops API. Let it be for the moment. Signed-off-by: -
Cédric Le Goater authored
This makes ihandle 64bit friendly. Signed-off-by:
-
Cédric Le Goater authored
This patch fixes 64bit compile warnings and updates the wrapper code to converge the kernel code in prom_init. Signed-off-by:
-
Cédric Le Goater authored
This is mostly useful to make to the boot wrapper code closer with the kernel code in prom_init. Signed-off-by:
-
Cédric Le Goater authored
Values will need to be byte-swapped when calling prom (big endian) from a little endian boot wrapper. Signed-off-by:
-
Cédric Le Goater authored
This patch updates the wrapper code to converge with the kernel code in prom_init. Signed-off-by:
-
Cédric Le Goater authored
This patch fixes warnings when the wrapper is compiled in 64bit and updates the boot wrapper code related to prom to converge with the kernel code in prom_init. This should make the review of changes easier. The kernel has a different number of possible arguments (10) when entering prom. There does not seem to be any good reason to have 12 in the wrapper, so the patch changes this value to args[10] in the prom_args struct. Signed-off-by:
-
Cédric Le Goater authored
When the boot wrapper is compiled in 64bit, there is no need to use __div64_32. Signed-off-by:
-
Gavin Shan authored
Function early_init_dt_scan_fw_dump() is called to scan the device tree for fdump properties under node "rtas". Any one of them is invalid, we can stop scanning the device tree early by returning "1". It would save a bit time during boot. Signed-off-by:
Gavin Shan <gwshan@linux.vnet.ibm.com> Signed-off-by:
-
Gavin Shan authored
If the PE contains single PCI function, "pe->pbus" would be NULL. It's not reliable to be used by pci_domain_nr(). We just grab the PCI domain number from the PCI host controller (struct pci_controller) instance. Signed-off-by:
-
Gavin Shan authored
In function pnv_pci_ioda2_setup_dma_pe(), the IOMMU table type is set to (TCE_PCI_SWINV_CREATE | TCE_PCI_SWINV_FREE) unconditionally. It was just set to TCE_PCI by pnv_pci_setup_iommu_table(). So the primary IOMMU table type (TCE_PCI) is lost. The patch fixes it. Also, pnv_pci_setup_iommu_table() already set "tbl->it_busno" to zero and we needn't do it again. The patch removes the redundant assignment. The patch also fixes similar issues in pnv_pci_ioda_setup_dma_pe(). Signed-off-by:
-
Gavin Shan authored
The patch intends to support fundamental reset on PLX downstream ports. If the PCI device matches any one of the internal table, which includes PLX vendor ID, bridge device ID, register offset for fundamental reset and bit, fundamental reset will be done accordingly. Otherwise, it will fail back to hot reset. Additional flag (EEH_DEV_FRESET) is introduced to record the last reset type on the PCI bridge. Signed-off-by:
-
Gavin Shan authored
When PCI_ERS_RESULT_CAN_RECOVER returned from device drivers, the EEH core should enable I/O and DMA for the affected PE. However, it was missed to have DMA enabled in eeh_handle_normal_event(). Besides, the frozen state of the affected PE should be cleared after successful recovery, but we didn't. The patch fixes both of the issues as above. Signed-off-by:
-
Gavin Shan authored
In the kdump scenario, the first kerenl doesn't shutdown PCI devices and the kdump kerenl clean PHB IODA table at the early probe time. That means the kdump kerenl can't support PCI transactions piled by the first kerenl. Otherwise, lots of EEH errors and frozen PEs will be detected. In order to avoid the EEH errors, the PHB is resetted to drop all PCI transaction from the first kerenl. Signed-off-by:
-
