aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* virtio: expose added descriptors immediately.Rusty Russell2012-01-121-6/+14
| | | | | | | | | | | | | | | | A virtio driver does virtqueue_add_buf() multiple times before finally calling virtqueue_kick(); previously we only exposed the added buffers in the virtqueue_kick() call. This means we don't need a memory barrier in virtqueue_add_buf(), but it reduces concurrency as the device (ie. host) can't see the buffers until the kick. In the unusual (but now possible) case where a driver does add_buf() and get_buf() without doing a kick, we do need to insert one before our counter wraps. Otherwise we could wrap num_added, and later on not realize that we have passed the marker where we should have kicked. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
* virtio: avoid modulus operation.Rusty Russell2012-01-121-4/+6
| | | | | | | Since we know vq->vring.num is a power of 2, modulus is lazy (it's asserted in vring_new_virtqueue()). Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
* virtio: support unlocked queue kickRusty Russell2012-01-122-12/+52
| | | | | | | | | | | | | | | | | | | | Based on patch by Christoph for virtio_blk speedup: Split virtqueue_kick to be able to do the actual notification outside the lock protecting the virtqueue. This patch was originally done by Stefan Hajnoczi, but I can't find the original one anymore and had to recreated it from memory. Pointers to the original or corrections for the commit message are welcome. Stefan's patch was here: https://github.com/stefanha/linux/commit/a6d06644e3a58e57a774e77d7dc34c4a5a2e7496 http://www.spinics.net/lists/linux-virtualization/msg14616.html Third time's the charm! Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
* virtio: rename virtqueue_add_buf_gfp to virtqueue_add_bufRusty Russell2012-01-1210-58/+44
| | | | | | | | | Remove wrapper functions. This makes the allocation type explicit in all callers; I used GPF_KERNEL where it seemed obvious, left it at GFP_ATOMIC otherwise. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Reviewed-by: Christoph Hellwig <hch@lst.de>
* virtio: document functions better.Rusty Russell2012-01-122-48/+91
| | | | | | | | | | | The old documentation is left over from when we used a structure with strategy pointers. And move the documentation to the C file as per kernel practice. Though I disagree... Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Reviewed-by: Christoph Hellwig <hch@lst.de>
* virtio-balloon: Trivial cleanupsSasha Levin2012-01-121-3/+3
| | | | | | | | | | Trivial changes to remove forgotten junk, format comments, and correct names. Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: virtualization@lists.linux-foundation.org Signed-off-by: Sasha Levin <levinsasha928@gmail.com> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
* virtio: harsher barriers for rpmsg.Rusty Russell2012-01-128-22/+35
| | | | | | | | | | | | | | | | | | We were cheating with our barriers; using the smp ones rather than the real device ones. That was fine, until rpmsg came along, which is used to talk to a real device (a non-SMP CPU). Unfortunately, just putting back the real barriers (reverting d57ed95d) causes a performance regression on virtio-pci. In particular, Amos reports netbench's TCP_RR over virtio_net CPU utilization increased up to 35% while throughput went down by up to 14%. By comparison, this branch is in the noise. Reference: https://lkml.org/lkml/2011/12/11/22 Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
* Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhostLinus Torvalds2012-01-1048-369/+124
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | lib: use generic pci_iomap on all architectures Many architectures don't want to pull in iomap.c, so they ended up duplicating pci_iomap from that file. That function isn't trivial, and we are going to modify it https://lkml.org/lkml/2011/11/14/183 so the duplication hurts. This reduces the scope of the problem significantly, by moving pci_iomap to a separate file and referencing that from all architectures. * tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost: alpha: drop pci_iomap/pci_iounmap from pci-noop.c mn10300: switch to GENERIC_PCI_IOMAP mn10300: add missing __iomap markers frv: switch to GENERIC_PCI_IOMAP tile: switch to GENERIC_PCI_IOMAP tile: don't panic on iomap sparc: switch to GENERIC_PCI_IOMAP sh: switch to GENERIC_PCI_IOMAP powerpc: switch to GENERIC_PCI_IOMAP parisc: switch to GENERIC_PCI_IOMAP mips: switch to GENERIC_PCI_IOMAP microblaze: switch to GENERIC_PCI_IOMAP arm: switch to GENERIC_PCI_IOMAP alpha: switch to GENERIC_PCI_IOMAP lib: add GENERIC_PCI_IOMAP lib: move GENERIC_IOMAP to lib/Kconfig Fix up trivial conflicts due to changes nearby in arch/{m68k,score}/Kconfig
| * alpha: drop pci_iomap/pci_iounmap from pci-noop.cMichael S. Tsirkin2011-12-041-12/+0
| | | | | | | | | | | | | | Jenkins disables PCI so asm-generic provides inline stubs for these, we don't need offline stubs as well. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * mn10300: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-12-044-33/+2
| | | | | | | | | | | | | | | | | | | | | | The pci_iomap variant that arch/mn10300/unit-asb2305/pci-iomap.c uses differs from the generic one in that it does not use ioremap_nocache for PCI addresses. However, it turns out that PCI addresses are automatically noncached, so switching to ioremap_nocache and to the generic implementation is safe. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * mn10300: add missing __iomap markersMichael S. Tsirkin2011-12-041-8/+8
| | | | | | | | | | | | | | MN10300's *ioremap*() collection lacks __iomem markers. Fix that. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * frv: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-12-044-31/+3
| | | | | | | | | | | | | | | | | | | | frv uses a version of pci_iomap that simply casts and returns back the start address. Looking closely, both ioremap and ioport_map seem to do this on this platform, so the generic pci_iomap will DTRT automatically. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * tile: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-12-043-22/+2
| | | | | | | | | | | | | | | | | | | | tile now has working stubs for ioport_map and ioremap such that the generic pci_iomap will DTRT: cast to pointer on memory and return NULL and log message on IO map. Switch it over to GENERIC_PCI_IOMAP. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * tile: don't panic on iomapMichael S. Tsirkin2011-12-041-1/+2
| | | | | | | | | | | | | | | | I think panic on iomap is there just for debugging. If we return NULL instead, the generic pci_iomap will DTRT so we don't need to roll our own. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * sparc: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-12-044-25/+3
| | | | | | | | | | | | | | | | sparc copied pci_iomap from generic code, probably to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * sh: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-282-23/+1
| | | | | | | | | | | | | | | | sh copied pci_iomap from generic code, probably to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * powerpc: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-282-19/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | powerpc copied pci_iomap from generic code, probably to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. The only difference is handling of nocache flag, that turns out to be done correctly by the generic code since arch/powerpc/include/asm/io.h defines ioremap_nocache same as ioremap. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * parisc: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-282-23/+1
| | | | | | | | | | | | | | | | parisc copied pci_iomap from generic code, probably to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * mips: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-282-26/+1
| | | | | | | | | | | | | | | | mips copied pci_iomap from generic code, probably to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * microblaze: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-282-19/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | microblaze copied pci_iomap from generic code, probably to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. The only difference is handling of nocache flag, that turns out to be done correctly by the generic code since arch/microblaze/include/asm/io.h defines ioremap_nocache same as ioremap. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * arm: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-283-22/+2
| | | | | | | | | | | | | | | | arm copied pci_iomap from generic code, probably to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * alpha: switch to GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-282-25/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | alpha copied pci_iomap from generic code to avoid pulling the rest of iomap.c in. Since that's in a separate file now, we can reuse the common implementation. The only difference is handling of nocache flag, that turns out to be done correctly by the generic code since arch/alpha/include/asm/io.h defines ioremap_nocache same as ioremap. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * lib: add GENERIC_PCI_IOMAPMichael S. Tsirkin2011-11-287-44/+85
| | | | | | | | | | | | | | | | | | Many architectures want a generic pci_iomap but not the rest of iomap.c. Split that to a separate .c file and add a new config symbol. select automatically by GENERIC_IOMAP. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
| * lib: move GENERIC_IOMAP to lib/KconfigMichael S. Tsirkin2011-11-2412-36/+10
| | | | | | | | | | | | | | | | | | define GENERIC_IOMAP in a central location instead of all architectures. This will be helpful for the follow-up patch which makes it select other configs. Code is also a bit shorter this way. Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
* | Merge tag 'for-linux-3.3-merge-window' of ↵Linus Torvalds2012-01-10125-5/+12989
|\ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | git://linux-c6x.org/git/projects/linux-c6x-upstreaming * tag 'for-linux-3.3-merge-window' of git://linux-c6x.org/git/projects/linux-c6x-upstreaming: (29 commits) C6X: replace tick_nohz_stop/restart_sched_tick calls C6X: add register_cpu call C6X: deal with memblock API changes C6X: fix timer64 initialization C6X: fix layout of EMIFA registers C6X: MAINTAINERS C6X: DSCR - Device State Configuration Registers C6X: EMIF - External Memory Interface C6X: general SoC support C6X: library code C6X: headers C6X: ptrace support C6X: loadable module support C6X: cache control C6X: clocks C6X: build infrastructure C6X: syscalls C6X: interrupt handling C6X: time management C6X: signal management ...
| * | C6X: replace tick_nohz_stop/restart_sched_tick callsMark Salter2012-01-081-2/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The following commits replaced the tick_nohz_{stop,restart}_sched_tick API with separate tick and rcu calls: 280f06774afedf849f0b34248ed6aff57d0f6908 2bbb6817c0ac1b5f2a68d720f364f98eeb1ac4fd 1268fbc746ea1cd279886a740dcbad4ba5232225 This patch replaces the C6X use of the old API with the newer interfaces. Signed-off-by: Mark Salter <msalter@redhat.com>
| * | C6X: add register_cpu callMark Salter2012-01-081-0/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | Commit ccbc60d3e19a1b6ae66ca0d89b3da02dde62088b requires CPU topology information even in !SMP cases. This requires C6X to add a call tp register_cpu() in order to avoid a panic at boot time. Signed-off-by: Mark Salter <msalter@redhat.com>
| * | C6X: deal with memblock API changesMark Salter2012-01-082-7/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Recent memblock related commits require the following C6X changes: * commit 24aa07882b672fff2da2f5c955759f0bd13d32d5 asm/memblock.h no longer required * commit 1440c4e2c918532f39131c3330fe2226e16be7b6 memblock_analyze() no longer needed to update total size * commit fe091c208a40299fba40e62292a610fb91e44b4e memblock_init() no longer needed Signed-off-by: Mark Salter <msalter@redhat.com>
| * | C6X: fix timer64 initializationMark Salter2012-01-081-1/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Some SoCs have a timer block enable controlled through the DSCR registers. There is a problem in the timer64 driver initialization where the code accesses a timer register to get the divisor used to calculate timer clock rate. If the timer block has not been enabled when this register read takes place, an exception is generated. This patch makes sure that the timer block is enabled before accessing the registers. Signed-off-by: Mark Salter <msalter@redhat.com>
| * | C6X: fix layout of EMIFA registersMark Salter2012-01-081-1/+0
| | | | | | | | | | | | Signed-off-by: Mark Salter <msalter@redhat.com>
| * | C6X: MAINTAINERSMark Salter2011-10-061-0/+8
| | | | | | | | | | | | | | | | | | Signed-off-by: Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: DSCR - Device State Configuration RegistersMark Salter2011-10-062-0/+632
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | All SoCs provide an area of device configuration registers called the DSCR. The location of specific registers as well as their use varies considerably from implementation to implementation. Rather than having to rely on additional SoC-specific DSCR code for each new supported SoC, this code generalize things as much as possible using device tree properties. Initialization must take place early on (setup_arch time) in case the event timer device needs to be enable via the DSCR. Signed-off-by: Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: EMIF - External Memory InterfaceMark Salter2011-10-061-0/+88
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Several SoC parts provide a simple bridge to support external memory mapped devices. This code probes the device tree for an EMIF node and sets up the bridge registers if such a node is found. Beyond initial set up, there is no further need to access the bridge control registers. External devices on the bus are accessed through their MMIO registers using suitable drivers. The bridge hardware does provide for timeout and other error interrupts, but these are not yet supported. Signed-off-by: Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: general SoC supportMark Salter2011-10-062-0/+126
| | | | | | | | | | | | | | | | | | | | | | | | | | | This patch provides a soc_ops struct which provides hooks for SoC functionality which doesn't fit well into other places. Signed-off-by: Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: library codeAurelien Jacquiot2011-10-0619-0/+1349
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: headersAurelien Jacquiot2011-10-0623-0/+1178
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: ptrace supportAurelien Jacquiot2011-10-062-0/+361
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: loadable module supportAurelien Jacquiot2011-10-063-0/+222
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: cache controlAurelien Jacquiot2011-10-063-0/+600
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: clocksMark Salter2011-10-064-0/+1018
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The C6X SoCs contain several PLL controllers each with up to 16 clock outputs feeding into the cores or peripheral clock domains. The hardware is very similar to arm/mach-davinci clocks. This is still a work in progress which needs to be updated once device tree clock binding changes shake out. Signed-off-by: Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: build infrastructureAurelien Jacquiot2011-10-064-0/+1385
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: syscallsAurelien Jacquiot2011-10-063-0/+155
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de> [msalter@redhat.com: add include of linux/module.h to sys_c6x.c] Signed-off-by: Mark Salter <msalter@redhat.com>
| * | C6X: interrupt handlingAurelien Jacquiot2011-10-066-0/+1480
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: time managementAurelien Jacquiot2011-10-064-0/+340
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: signal managementAurelien Jacquiot2011-10-063-0/+474
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: process managementAurelien Jacquiot2011-10-064-0/+590
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: memory management and DMA supportAurelien Jacquiot2011-10-064-0/+500
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> The C6X architecture currently lacks an MMU so memory management is relatively simple. There is no bus snooping between L2 and main memory but coherent DMA memory is supported by making regions of main memory uncached. If such a region is desired, it can be specified on the commandline with a "memdma=" argument. Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: devicetree supportMark Salter2011-10-0617-0/+1087
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is the basic devicetree support for C6X. Currently, four boards are supported. Each one uses a different SoC part. Two of the four supported SoCs are multicore. One with 3 cores and the other with 6 cores. There is no coherency between the core-level caches, so SMP is not an option. It is possible to run separate kernel instances on the various cores. There is currently no C6X bootloader support for device trees so we build in the DTB for now. There are some interesting twists to the hardware which are of note for device tree support. Each core has its own interrupt controller which is controlled by special purpose core registers. This core controller provides 12 general purpose prioritized interrupt sources. Each core is contained within a hardware "module" which provides L1 and L2 caches, power control, and another interrupt controller which cascades into the core interrupt controller. These core module functions are controlled by memory mapped registers. The addresses for these registers are the same for each core. That is, when coreN accesses a module-level MMIO register at a given address, it accesses the register for coreN even though other cores would use the same address to access the register in the module containing those cores. Other hardware modules (timers, enet, etc) which are memory mapped can be accessed by all cores. The timers need some further explanation for multicore SoCs. Even though all timer control registers are visible to all cores, interrupt routing or other considerations may make a given timer more suitable for use by a core than some other timer. Because of this and the desire to have the same image run on more than one core, the timer nodes have a "ti,core-mask" property which is used by the driver to scan for a suitable timer to use. Signed-off-by: Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: early boot codeAurelien Jacquiot2011-10-063-0/+663
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> This patch provides the early boot code for C6X architecture. There is a 16 entry vector table which is used to direct reset and interrupt events. The vector table entries contain a small amount of code (maximum of 8 opcodes) which simply branches to the actual event handling code. The head.S code simply clears BSS, setups up a few control registers, and calls machine_init followed by start_kernel. The machine_init code in setup.c does the early flat tree parsing (memory, commandline, etc). At setup_arch time, the code does the usual memory setup and minimally scans the devicetree for any needed information. Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
| * | C6X: build infrastructureAurelien Jacquiot2011-10-0614-0/+701
| | | | | | | | | | | | | | | | | | | | | | | | | | | Original port to early 2.6 kernel using TI COFF toolchain. Brought up to date by Mark Salter <msalter@redhat.com> Signed-off-by: Aurelien Jacquiot <a-jacquiot@ti.com> Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de>
generated by cgit v1.1 at 2024-12-27 17:39:04 +0000