- 29 Dec, 2017 40 commits
-
-
Wanpeng Li authored
commit d73235d1 upstream. *** Guest State *** CR0: actual=0x0000000000000030, shadow=0x0000000060000010, gh_mask=fffffffffffffff7 CR4: actual=0x0000000000002050, shadow=0x0000000000000000, gh_mask=ffffffffffffe871 CR3 = 0x00000000fffbc000 RSP = 0x0000000000000000 RIP = 0x0000000000000000 RFLAGS=0x00000000 DR7 = 0x0000000000000400 ^^^^^^^^^^ The failed vmentry is triggered by the following testcase when ept=Y: #include <unistd.h> #include <sys/syscall.h> #include <string.h> #include <stdint.h> #include <linux/kvm.h> #include <fcntl.h> #include <sys/ioctl.h> long r[5]; int main() { r[2] = open("/dev/kvm", O_RDONLY); r[3] = ioctl(r[2], KVM_CREATE_VM, 0); r[4] = ioctl(r[3], KVM_CREATE_VCPU, 7); struct kvm_regs regs = { .rflags = 0, }; ioctl(r[4], KVM_SET_REGS, ®s); ioctl(r[4], KVM_RUN, 0); } X86 RFLAGS bit 1 is fixed set, userspace can simply clearing bit 1 of RFLAGS with KVM_SET_REGS ioctl which results in vmentry fails. This patch fixes it by oring X86_EFLAGS_FIXED during ioctl. Suggested-by:
Jim Mattson <jmattson@google.com> Reviewed-by:
David Hildenbrand <david@redhat.com> Reviewed-by:
Quan Xu <quan.xu0@gmail.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Jim Mattson <jmattson@google.com> Signed-off-by:
Wanpeng Li <wanpeng.li@hotmail.com> Signed-off-by:
Paolo Bonzini <pbonzini@redhat.com> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Wanpeng Li authored
commit ed52870f upstream. The below test case can cause infinite loop in kvm when ept=0. #include <unistd.h> #include <sys/syscall.h> #include <string.h> #include <stdint.h> #include <linux/kvm.h> #include <fcntl.h> #include <sys/ioctl.h> long r[5]; int main() { r[2] = open("/dev/kvm", O_RDONLY); r[3] = ioctl(r[2], KVM_CREATE_VM, 0); r[4] = ioctl(r[3], KVM_CREATE_VCPU, 7); ioctl(r[4], KVM_RUN, 0); } It doesn't setup the memory regions, mmu_alloc_shadow/direct_roots() in kvm return 1 when kvm fails to allocate root page table which can result in beblow infinite loop: vcpu_run() { for (;;) { r = vcpu_enter_guest()::kvm_mmu_reload() returns 1 if (r <= 0) break; if (need_resched()) cond_resched(); } } This patch fixes it by returning -ENOSPC when there is no available kvm mmu page for root page table. Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Fixes: 26eeb53c (KVM: MMU: Bail out immediately if there is no available mmu page) Signed-off-by:
-
Laurent Vivier authored
commit 7333b5ac upstream. When we migrate a VM from a POWER8 host (XICS) to a POWER9 host (XICS-on-XIVE), we have an error: qemu-kvm: Unable to restore KVM interrupt controller state \ (0xff000000) for CPU 0: Invalid argument This is because kvmppc_xics_set_icp() checks the new state is internaly consistent, and especially: ... 1129 if (xisr == 0) { 1130 if (pending_pri != 0xff) 1131 return -EINVAL; ... On the other side, kvmppc_xive_get_icp() doesn't set neither the pending_pri value, nor the xisr value (set to 0) (and kvmppc_xive_set_icp() ignores the pending_pri value) As xisr is 0, pending_pri must be set to 0xff. Fixes: 5af50993 ("KVM: PPC: Book3S HV: Native usage of the XIVE interrupt controller") Signed-off-by:
Laurent Vivier <lvivier@redhat.com> Acked-by:
Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by:
Michael Ellerman <mpe@ellerman.id.au> Signed-off-by:
-
Cédric Le Goater authored
commit dc1c4165 upstream. When restoring a pending interrupt, we are setting the Q bit to force a retrigger in xive_finish_unmask(). But we also need to force an EOI in this case to reach the same initial state : P=1, Q=0. This can be done by not setting 'old_p' for pending interrupts which will inform xive_finish_unmask() that an EOI needs to be sent. Fixes: 5af50993 ("KVM: PPC: Book3S HV: Native usage of the XIVE interrupt controller") Suggested-by:
Cédric Le Goater <clg@kaod.org> Reviewed-by:
-
Marc Zyngier authored
commit 7839c672 upstream. When we unmap the HYP memory, we try to be clever and unmap one PGD at a time. If we start with a non-PGD aligned address and try to unmap a whole PGD, things go horribly wrong in unmap_hyp_range (addr and end can never match, and it all goes really badly as we keep incrementing pgd and parse random memory as page tables...). The obvious fix is to let unmap_hyp_range do what it does best, which is to iterate over a range. The size of the linear mapping, which begins at PAGE_OFFSET, can be easily calculated by subtracting PAGE_OFFSET form high_memory, because high_memory is defined as the linear map address of the last byte of DRAM, plus one. The size of the vmalloc region is given trivially by VMALLOC_END - VMALLOC_START. Reported-by:
Andre Przywara <andre.przywara@arm.com> Tested-by:
Christoffer Dall <christoffer.dall@linaro.org> Signed-off-by:
Marc Zyngier <marc.zyngier@arm.com> Signed-off-by:
-
Julien Thierry authored
commit bfe766cf upstream. When VHE is not present, KVM needs to save and restores PMSCR_EL1 when possible. If SPE is used by the host, value of PMSCR_EL1 cannot be saved for the guest. If the host starts using SPE between two save+restore on the same vcpu, restore will write the value of PMSCR_EL1 read during the first save. Make sure __debug_save_spe_nvhe clears the value of the saved PMSCR_EL1 when the guest cannot use SPE. Signed-off-by:
Julien Thierry <julien.thierry@arm.com> Cc: Christoffer Dall <christoffer.dall@linaro.org> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Reviewed-by:
Will Deacon <will.deacon@arm.com> Reviewed-by:
-
Mika Westerberg authored
commit d2b3c353 upstream. Guenter Roeck reported an interrupt storm on a prototype system which is based on Cyan Chromebook. The root cause turned out to be a incorrectly configured pin that triggers spurious interrupts. This will be fixed in coreboot but currently we need to prevent the interrupt storm from happening by masking all interrupts (but not GPEs) on those systems. Link: https://bugzilla.kernel.org/show_bug.cgi?id=197953 Fixes: bcb48cca ("pinctrl: cherryview: Do not mask all interrupts in probe") Reported-and-tested-by:
Guenter Roeck <linux@roeck-us.net> Reported-by:
Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by:
Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by:
Linus Walleij <linus.walleij@linaro.org> Signed-off-by:
-
Maxime Chevallier authored
commit 251c201b upstream. The Armada 3700 SPI controller has 2 ranges of prescaler coefficients. One ranging from 0 to 15 by steps of 1, and one ranging from 0 to 30 by steps of 2. This commit fixes the prescaler coefficients that are over 15 so that it uses the correct range of values. The prescaling coefficient is rounded to the upper value if it is odd. This was tested on Espressobin with spidev and a locigal analyser. Signed-off-by:
Maxime Chevallier <maxime.chevallier@smile.fr> Signed-off-by:
Mark Brown <broonie@kernel.org> Signed-off-by:
-
Ricardo Ribalda Delgado authored
commit 5a1314fa upstream. When the core is configured in C_SPI_MODE > 0, it integrates a lookup table that automatically configures the core in dual or quad mode based on the command (first byte on the tx fifo). Unfortunately, that list mode_?_memoy_*.mif does not contain all the supported commands by the flash. Since 4.14 spi-nor automatically tries to probe the flash using SFDP (command 0x5a), and that command is not part of the list_mode table. Whit the right combination of C_SPI_MODE and C_SPI_MEMORY this leads into a stall that can only be recovered with a soft rest. This patch detects this kind of stall and returns -EIO to the caller on those commands. spi-nor can handle this error properly: m25p80 spi0.0: Detected stall. Check C_SPI_MODE and C_SPI_MEMORY. 0x21 0x2404 m25p80 spi0.0: SPI transfer failed: -5 spi_master spi0: failed to transfer one message from queue m25p80 spi0.0: s25sl064p (8192 Kbytes) Signed-off-by:
Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com> Signed-off-by:
-
John David Anglin authored
commit 9352aead upstream. This reverts commit 5c38602d. Interrupts can't be enabled early because the register saves are done on the thread stack prior to switching to the IRQ stack. This caused stack overflows and the thread stack needed increasing to 32k. Even then, stack overflows still occasionally occurred. Background: Even with a 32 kB thread stack, I have seen instances where the thread stack overflowed on the mx3210 buildd. Detection of stack overflow only occurs when we have an external interrupt. When an external interrupt occurs, we switch to the thread stack if we are not already on a kernel stack. Then, registers and specials are saved to the kernel stack. The bug occurs in intr_return where interrupts are reenabled prior to returning from the interrupt. This was done incase we need to schedule or deliver signals. However, it introduces the possibility that multiple external interrupts may occur on the thread stack and cause a stack overflow. These might not be detected and cause the kernel to misbehave in random ways. This patch changes the code back to only reenable interrupts when we are going to schedule or deliver signals. As a result, we generally return from an interrupt before reenabling interrupts. This minimizes the growth of the thread stack. Fixes: 5c38602d ("parisc: Re-enable interrupts early") Signed-off-by:
John David Anglin <dave.anglin@bell.net> Signed-off-by:
Helge Deller <deller@gmx.de> Signed-off-by:
-
Helge Deller authored
commit bcf3f175 upstream. Diva GSP card has built-in serial AUX port and ATI graphic card which simply don't work and which both don't have external connectors. User Guides even mention that those devices shouldn't be used. So, prevent that Linux drivers try to enable those devices. Signed-off-by:
-
Helge Deller authored
commit 203c110b upstream. Static analysis tools complain that we intended to have curly braces around this indent block. In this case this assumption is wrong, so fix the indenting. Fixes: 2f3c7b81 ("parisc: Add core code for self-extracting kernel") Reported-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
Helge Deller authored
commit 0ed9d3de upstream. The os_hpmc_size variable sometimes wasn't aligned at word boundary and thus triggered the unaligned fault handler at startup. Fix it by aligning it properly. Signed-off-by:
-
Shaohua Li authored
commit 111be883 upstream. If a bio is throttled and split after throttling, the bio could be resubmited and enters the throttling again. This will cause part of the bio to be charged multiple times. If the cgroup has an IO limit, the double charge will significantly harm the performance. The bio split becomes quite common after arbitrary bio size change. To fix this, we always set the BIO_THROTTLED flag if a bio is throttled. If the bio is cloned/split, we copy the flag to new bio too to avoid a double charge. However, cloned bio could be directed to a new disk, keeping the flag be a problem. The observation is we always set new disk for the bio in this case, so we can clear the flag in bio_set_dev(). This issue exists for a long time, arbitrary bio size change just makes it worse, so this should go into stable at least since v4.2. V1-> V2: Not add extra field in bio based on discussion with Tejun Cc: Vivek Goyal <vgoyal@redhat.com> Acked-by:
Tejun Heo <tj@kernel.org> Signed-off-by:
Shaohua Li <shli@fb.com> Signed-off-by:
Jens Axboe <axboe@kernel.dk> Signed-off-by:
-
Jens Axboe authored
commit 4ccafe03 upstream. A previous change blindly added massive alignment to the call_single_data structure in struct request. This ballooned it in size from 296 to 320 bytes on my setup, for no valid reason at all. Use the unaligned struct __call_single_data variant instead. Fixes: 966a9671 ("smp: Avoid using two cache lines for struct call_single_data") Signed-off-by:
-
Rafael J. Wysocki authored
commit 5839ee73 upstream. It is incorrect to call pci_restore_state() for devices in low-power states (D1-D3), as that involves the restoration of MSI setup which requires MMIO to be operational and that is only the case in D0. However, pci_pm_thaw_noirq() may do that if the driver's "freeze" callbacks put the device into a low-power state, so fix it by making it force devices into D0 via pci_set_power_state() instead of trying to "update" their power state which is pointless. Fixes: e60514bd (PCI/PM: Restore the status of PCI devices across hibernation) Reported-by:
Thomas Gleixner <tglx@linutronix.de> Reported-by:
Maarten Lankhorst <dev@mblankhorst.nl> Tested-by:
Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by:
Bjorn Helgaas <bhelgaas@google.com> Signed-off-by:
-
Takashi Iwai authored
commit 5a15f289 upstream. The commit 89b89d12 ("ALSA: usb-audio: Add check return value for usb_string()") added the check of the return value from snd_usb_copy_string_desc(), which is correct per se, but it introduced a regression. In the original code, either the "Clock Source", "Playback Source" or "Capture Source" suffix is added after the terminal string, while the commit changed it to add the suffix only when get_term_name() is failing. It ended up with an incorrect ctl name like "PCM" instead of "PCM Capture Source". Also, even the original code has a similar bug: when the ctl name is generated from snd_usb_copy_string_desc() for the given iSelector, it also doesn't put the suffix. This patch addresses these issues: the suffix is added always when no static mapping is found. Also the patch tries to put more comments and cleans up the if/else block for better readability in order to avoid the same pitfall again. Fixes: 89b89d12 ("ALSA: usb-audio: Add check return value for usb_string()") Reported-and-tested-by:
Mauro Santos <registo.mailling@gmail.com> Signed-off-by:
Takashi Iwai <tiwai@suse.de> Signed-off-by:
-
Jussi Laako authored
commit 866f7ed7 upstream. Adds VID:PID of Esoteric D-05X to the TEAC device id's. Renames the is_teac_50X_dac() function to is_teac_dsd_dac() to cover broader device family from the same corporation sharing the same USB audio implementation. Signed-off-by:
Jussi Laako <jussi@sonarnerd.net> Signed-off-by:
-
Guneshwor Singh authored
commit 2b4584d0 upstream. Cannonlake HDMI codec has the same nid as Geminilake. This adds the codec entry for it. Signed-off-by:
Guneshwor Singh <guneshwor.o.singh@intel.com> Signed-off-by:
-
Kailang Yang authored
commit 92266651 upstream. Dell AIO had LineOut jack. Add LineOut verb into this patch. [ Additional notes: the ALC274 codec seems requiring the fixed pin / DAC connections for HP / line-out pins for enabling EQ for speakers; i.e. the HP / LO pins expect to be connected with NID 0x03 while keeping the speaker with NID 0x02. However, by adding a new line-out pin, the auto-parser assigns the NID 0x02 for HP/LO pins as primary outputs. As an easy workaround, we provide the preferred_pairs[] to map forcibly for these pins. -- tiwai ] Fixes: 75ee94b2 ("ALSA: hda - fix headset mic problem for Dell machines with alc274") Signed-off-by:
Kailang Yang <kailang@realtek.com> Signed-off-by:
-
Takashi Iwai authored
commit c1cfd902 upstream. The rawmidi also allows to obtaining the information via ioctl of ctl API. It means that user can issue an ioctl to the rawmidi device even when it's being removed as long as the control device is present. Although the code has some protection via the global register_mutex, its range is limited to the search of the corresponding rawmidi object, and the mutex is already unlocked at accessing the rawmidi object. This may lead to a use-after-free. For avoiding it, this patch widens the application of register_mutex to the whole snd_rawmidi_info_select() function. We have another mutex per rawmidi object, but this operation isn't very hot path, so it shouldn't matter from the performance POV. Signed-off-by:
-
Johan Hovold authored
commit 85e9b13c upstream. Fix child-node lookup during probe, which ended up searching the whole device tree depth-first starting at the parent rather than just matching on its children. To make things worse, the parent node was prematurely freed, while the child node was leaked. Note that the CONFIG_OF compile guard can be removed as of_get_child_by_name() provides a !CONFIG_OF implementation which always fails. Fixes: 37e13cec ("mfd: Add support for Device Tree to twl6040") Fixes: ca2cad6a ("mfd: Fix twl6040 build failure") Signed-off-by:
Johan Hovold <johan@kernel.org> Acked-by:
Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by:
Lee Jones <lee.jones@linaro.org> Signed-off-by:
-
Johan Hovold authored
commit 0a423772 upstream. A helper purported to look up a child node based on its name was using the wrong of-helper and ended up prematurely freeing the parent of-node while leaking any matching node. To make things worse, any matching node would not even necessarily be a child node as the whole device tree was searched depth-first starting at the parent. Fixes: 019a7e6b ("mfd: twl4030-audio: Add DT support") Signed-off-by:
-
Jon Hunter authored
commit 15d83748 upstream. On the Tegra124 Nyan-Big chromebook the very first SPI message sent to the EC is failing. The Tegra SPI driver configures the SPI chip-selects to be active-high by default (and always has for many years). The EC SPI requires an active-low chip-select and so the Tegra chip-select is reconfigured to be active-low when the EC SPI driver calls spi_setup(). The problem is that if the first SPI message to the EC is sent too soon after reconfiguring the SPI chip-select, it fails. The EC SPI driver prevents back-to-back SPI messages being sent too soon by keeping track of the time the last transfer was sent via the variable 'last_transfer_ns'. To prevent the very first transfer being sent too soon, initialise the 'last_transfer_ns' variable after calling spi_setup() and before sending the first SPI message. Signed-off-by:
Jon Hunter <jonathanh@nvidia.com> Reviewed-by:
Brian Norris <briannorris@chromium.org> Reviewed-by:
Douglas Anderson <dianders@chromium.org> Acked-by:
Benson Leung <bleung@chromium.org> Signed-off-by:
-
Stephan Mueller authored
commit d53c5135 upstream. When invoking an asynchronous cipher operation, the invocation of the callback may be performed before the subsequent operations in the initial code path are invoked. The callback deletes the cipher request data structure which implies that after the invocation of the asynchronous cipher operation, this data structure must not be accessed any more. The setting of the return code size with the request data structure must therefore be moved before the invocation of the asynchronous cipher operation. Fixes: e870456d ("crypto: algif_skcipher - overhaul memory management") Fixes: d887c52d ("crypto: algif_aead - overhaul memory management") Reported-by:
syzbot <syzkaller@googlegroups.com> Signed-off-by:
Stephan Mueller <smueller@chronox.de> Acked-by:
Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by:
Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by:
-
Stephan Mueller authored
commit 11edb555 upstream. The wait for data is a non-atomic operation that can sleep and therefore potentially release the socket lock. The release of the socket lock allows another thread to modify the context data structure. The waiting operation for new data therefore must be called at the beginning of recvmsg. This prevents a race condition where checks of the members of the context data structure are performed by recvmsg while there is a potential for modification of these values. Fixes: e870456d ("crypto: algif_skcipher - overhaul memory management") Fixes: d887c52d ("crypto: algif_aead - overhaul memory management") Reported-by:
-
Sebastian Andrzej Siewior authored
commit 9abffc6f upstream. mcryptd_enqueue_request() grabs the per-CPU queue struct and protects access to it with disabled preemption. Then it schedules a worker on the same CPU. The worker in mcryptd_queue_worker() guards access to the same per-CPU variable with disabled preemption. If we take CPU-hotplug into account then it is possible that between queue_work_on() and the actual invocation of the worker the CPU goes down and the worker will be scheduled on _another_ CPU. And here the preempt_disable() protection does not work anymore. The easiest thing is to add a spin_lock() to guard access to the list. Another detail: mcryptd_queue_worker() is not processing more than MCRYPTD_BATCH invocation in a row. If there are still items left, then it will invoke queue_work() to proceed with more later. *I* would suggest to simply drop that check because it does not use a system workqueue and the workqueue is already marked as "CPU_INTENSIVE". And if preemption is required then the scheduler should do it. However if queue_work() is used then the work item is marked as CPU unbound. That means it will try to run on the local CPU but it may run on another CPU as well. Especially with CONFIG_DEBUG_WQ_FORCE_RR_CPU=y. Again, the preempt_disable() won't work here but lock which was introduced will help. In order to keep work-item on the local CPU (and avoid RR) I changed it to queue_work_on(). Signed-off-by:
Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by:
-
Eric Biggers authored
commit 2b4f27c3 upstream. All the ChaCha20 algorithms as well as the ARM bit-sliced AES-XTS algorithms call skcipher_walk_virt(), then access the IV (walk.iv) before checking whether any bytes need to be processed (walk.nbytes). But if the input is empty, then skcipher_walk_virt() doesn't set the IV, and the algorithms crash trying to use the uninitialized IV pointer. Fix it by setting the IV earlier in skcipher_walk_virt(). Also fix it for the AEAD walk functions. This isn't a perfect solution because we can't actually align the IV to ->cra_alignmask unless there are bytes to process, for one because the temporary buffer for the aligned IV is freed by skcipher_walk_done(), which is only called when there are bytes to process. Thus, algorithms that require aligned IVs will still need to avoid accessing the IV when walk.nbytes == 0. Still, many algorithms/architectures are fine with IVs having any alignment, and even for those that aren't, a misaligned pointer bug is much less severe than an uninitialized pointer bug. This change also matches the behavior of the older blkcipher_walk API. Fixes: 0cabf2af ("crypto: skcipher - Fix crash on zero-length input") Reported-by:
Eric Biggers <ebiggers@google.com> Signed-off-by:
-
Dan Williams authored
commit adf68957 upstream. Integration testing with a BIOS that generates injected health event notifications fails to communicate those events to userspace. The nfit driver neglects to link the ACPI DIMM device with the necessary driver data so acpi_nvdimm_notify() fails this lookup: nfit_mem = dev_get_drvdata(dev); if (nfit_mem && nfit_mem->flags_attr) sysfs_notify_dirent(nfit_mem->flags_attr); Add the necessary linkage when installing the notification handler and clean it up when the nfit driver instance is torn down. Cc: Toshi Kani <toshi.kani@hpe.com> Cc: Vishal Verma <vishal.l.verma@intel.com> Fixes: ba9c8dd3 ("acpi, nfit: add dimm device notification support") Reported-by:
Daniel Osawa <daniel.k.osawa@intel.com> Tested-by:
Dan Williams <dan.j.williams@intel.com> Signed-off-by:
-
Takashi Iwai authored
commit bb82e0b4 upstream. The commit f6f82851 ("pstore: pass allocated memory region back to caller") changed the check of the return value from erst_read() in erst_reader() in the following way: if (len == -ENOENT) goto skip; - else if (len < 0) { - rc = -1; + else if (len < sizeof(*rcd)) { + rc = -EIO; goto out; This introduced another bug: since the comparison with sizeof() is cast to unsigned, a negative len value doesn't hit any longer. As a result, when an error is returned from erst_read(), the code falls through, and it may eventually lead to some weird thing like memory corruption. This patch adds the negative error value check more explicitly for addressing the issue. Fixes: f6f82851 (pstore: pass allocated memory region back to caller) Tested-by:
Jerry Tang <jtang@suse.com> Signed-off-by:
Kees Cook <keescook@chromium.org> Reviewed-by:
Borislav Petkov <bp@suse.de> Signed-off-by:
-
Thomas Gleixner authored
commit f6c4fd50 upstream. The loop which populates the CPU entry area PMDs can wrap around on 32bit machines when the number of CPUs is small. It worked wonderful for NR_CPUS=64 for whatever reason and the moron who wrote that code did not bother to test it with !SMP. Check for the wraparound to fix it. Fixes: 92a0f81d ("x86/cpu_entry_area: Move it out of the fixmap") Reported-by:
kernel test robot <fengguang.wu@intel.com> Signed-off-by:
Borislav Petkov <bp@alien8.de> Signed-off-by:
-
Thomas Gleixner authored
commit 613e396b upstream. init_espfix_bsp() needs to be invoked before the page table isolation initialization. Move it into mm_init() which is the place where pti_init() will be added. While at it get rid of the #ifdeffery and provide proper stub functions. Signed-off-by:
Ingo Molnar <mingo@kernel.org> Signed-off-by:
-
Thomas Gleixner authored
commit 92a0f81d upstream. Put the cpu_entry_area into a separate P4D entry. The fixmap gets too big and 0-day already hit a case where the fixmap PTEs were cleared by cleanup_highmap(). Aside of that the fixmap API is a pain as it's all backwards. Signed-off-by:
-
Thomas Gleixner authored
commit ed1bbc40 upstream. Separate the cpu_entry_area code out of cpu/common.c and the fixmap. Signed-off-by:
-
Peter Zijlstra authored
commit 1a3b0cae upstream. Unclutter tlbflush.h a little. Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Brian Gerst <brgerst@gmail.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David Laight <David.Laight@aculab.com> Cc: Denys Vlasenko <dvlasenk@redhat.com> Cc: Eduardo Valentin <eduval@amazon.com> Cc: Greg KH <gregkh@linuxfoundation.org> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Juergen Gross <jgross@suse.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Will Deacon <will.deacon@arm.com> Cc: aliguori@amazon.com Cc: daniel.gruss@iaik.tugraz.at Cc: hughd@google.com Cc: keescook@google.com Cc: linux-mm@kvack.org Signed-off-by:
-
Dave Hansen authored
commit dd95f1a4 upstream. There are effectively two ASID types: 1. The one stored in the mmu_context that goes from 0..5 2. The one programmed into the hardware that goes from 1..6 This consolidates the locations where converting between the two (by doing a +1) to a single place which gives us a nice place to comment. PAGE_TABLE_ISOLATION will also need to, given an ASID, know which hardware ASID to flush for the userspace mapping. Signed-off-by:
Dave Hansen <dave.hansen@linux.intel.com> Signed-off-by:
-
Dave Hansen authored
commit cb0a9144 upstream. First, it's nice to remove the magic numbers. Second, PAGE_TABLE_ISOLATION is going to consume half of the available ASID space. The space is currently unused, but add a comment to spell out this new restriction. Signed-off-by:
-
Dave Hansen authored
commit 50fb83a6 upstream. For flushing the TLB, the ASID which has been programmed into the hardware must be known. That differs from what is in 'cpu_tlbstate'. Add functions to transform the 'cpu_tlbstate' values into to the one programmed into the hardware (CR3). It's not easy to include mmu_context.h into tlbflush.h, so just move the CR3 building over to tlbflush.h. Signed-off-by:
-
Peter Zijlstra authored
commit 3f67af51 upstream. Per popular request.. Signed-off-by:
-
Peter Zijlstra authored
commit b5fc6d94 upstream. atomic64_inc_return() already implies smp_mb() before and after. Signed-off-by:
-
