- linux-morello - op-lists.linaro.org

by carsten.haitzler＠foss.arm.com

From: Carsten Haitzler <carsten.haitzler(a)arm.com> This series starts to enable purecap support for drm ioctls. This series enables all the libdrm tests (tested and working). You will also need the purecap libdrm port as well to complement this. Carsten Haitzler (3): drm: Fix copy to/from user so that caps transport in the region drm: Fix purecap vblank handling drm: fix up purecap handling of all iotcls in libdrm test tools drivers/gpu/drm/drm_atomic_uapi.c | 104 ++++++++++++++++++------ drivers/gpu/drm/drm_connector.c | 108 +++++++++++++++++++------ drivers/gpu/drm/drm_file.c | 2 +- drivers/gpu/drm/drm_ioctl.c | 6 +- drivers/gpu/drm/drm_mode_config.c | 126 +++++++++++++++++++++--------- drivers/gpu/drm/drm_mode_object.c | 37 ++++++++- drivers/gpu/drm/drm_plane.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_property.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_vblank.c | 39 ++++++--- include/drm/drm_vblank.h | 23 ++++++ include/uapi/drm/drm.h | 6 +- include/uapi/drm/drm_mode.h | 42 +++++----- tools/include/uapi/drm/drm.h | 2 +- 13 files changed, 509 insertions(+), 164 deletions(-) -- 2.25.1

5 months

2
4
0 0

[PATCH v3] mmc: ioctl: Remove hard-coded MMC_IOC_{MULTI_}CMD values

by Akram Ahmad

The "mmc: Support capabilities in MMC_IOC_{MULTI_}CMD ioctls" patch previously hard-coded the ioctl command value in order to maintain compatibility with non-PCuABI ioctl handlers such as those found in block.c. This deviates from the standard approach which is to always define the ioctl command value in terms of the struct size (which may change), instead providing any fixes in the kernel via compat ioctl handlers. This patch therefore aligns the MMC ioctl handlers in block.c with the standard approach. Undoing the hard-coding in the header leaves the ioctl command value unchanged in all situations except in PCuABI where the structs sizes, and hence the ioctl command values, are now larger (the ioctl value remains unchanged in compat64 as well). The additional comments that were added in the aforementioned patch regarding the changing size of the structs is also removed in this patch. Signed-off-by: Akram Ahmad <Akram.Ahmad(a)arm.com> --- drivers/mmc/core/block.c | 64 ++++++++++++++++++++++++++++++++-- include/uapi/linux/mmc/ioctl.h | 22 +++--------- 2 files changed, 66 insertions(+), 20 deletions(-) diff --git a/drivers/mmc/core/block.c b/drivers/mmc/core/block.c index 17a1d24bdaed..ad63ba425d47 100644 --- a/drivers/mmc/core/block.c +++ b/drivers/mmc/core/block.c @@ -76,6 +76,15 @@ MODULE_ALIAS("mmc:block"); #define MMC_EXTRACT_INDEX_FROM_ARG(x) ((x & 0x00FF0000) >> 16) #define MMC_EXTRACT_VALUE_FROM_ARG(x) ((x & 0x0000FF00) >> 8) +/* + * The compat_mmc_ioc_{multi_}cmd structs have different sizes in PCuABI and in + * compat64, which affects the expected MMC_IOC_{MULTI_}CMD ioctl values. The + * following ioctl values reflect the sizes of the compat64 structs defined in + * this file. + */ +#define MMC_IOC_CMD_COMPAT _IOWR(MMC_BLOCK_MAJOR, 0, struct compat_mmc_ioc_cmd) +#define MMC_IOC_MULTI_CMD_COMPAT _IOWR(MMC_BLOCK_MAJOR, 1, struct compat_mmc_ioc_multi_cmd) + static DEFINE_MUTEX(block_mutex); /* @@ -891,7 +900,38 @@ static int mmc_blk_ioctl(struct block_device *bdev, blk_mode_t mode, static int mmc_blk_compat_ioctl(struct block_device *bdev, blk_mode_t mode, unsigned int cmd, unsigned long arg) { - return mmc_blk_ioctl(bdev, mode, cmd, (user_uintptr_t) compat_ptr(arg)); + struct mmc_blk_data *md; + int ret; + void __user *uarg = compat_ptr(arg); + + switch (cmd) { + case MMC_IOC_CMD_COMPAT: + ret = mmc_blk_check_blkdev(bdev); + if (ret) + return ret; + md = mmc_blk_get(bdev->bd_disk); + if (!md) + return -EINVAL; + ret = mmc_blk_ioctl_cmd(md, + uarg, + NULL); + mmc_blk_put(md); + return ret; + case MMC_IOC_MULTI_CMD_COMPAT: + ret = mmc_blk_check_blkdev(bdev); + if (ret) + return ret; + md = mmc_blk_get(bdev->bd_disk); + if (!md) + return -EINVAL; + ret = mmc_blk_ioctl_multi_cmd(md, + uarg, + NULL); + mmc_blk_put(md); + return ret; + default: + return -EINVAL; + } } #endif @@ -2700,7 +2740,27 @@ static long mmc_rpmb_ioctl(struct file *filp, unsigned int cmd, static long mmc_rpmb_ioctl_compat(struct file *filp, unsigned int cmd, unsigned long arg) { - return mmc_rpmb_ioctl(filp, cmd, (user_uintptr_t)compat_ptr(arg)); + struct mmc_rpmb_data *rpmb = filp->private_data; + int ret; + void __user *uarg = compat_ptr(arg); + + switch (cmd) { + case MMC_IOC_CMD_COMPAT: + ret = mmc_blk_ioctl_cmd(rpmb->md, + uarg, + rpmb); + break; + case MMC_IOC_MULTI_CMD_COMPAT: + ret = mmc_blk_ioctl_multi_cmd(rpmb->md, + uarg, + rpmb); + break; + default: + ret = -EINVAL; + break; + } + + return ret; } #endif diff --git a/include/uapi/linux/mmc/ioctl.h b/include/uapi/linux/mmc/ioctl.h index 9d34bfc49198..3ca3cd4785f7 100644 --- a/include/uapi/linux/mmc/ioctl.h +++ b/include/uapi/linux/mmc/ioctl.h @@ -64,29 +64,15 @@ struct mmc_ioc_multi_cmd { struct mmc_ioc_cmd cmds[]; }; -/* - * The size of struct mmc_ioc_cmd changes in PCuABI due to the use of - * __kernel_uintptr_t, which in turn modifies the command value for this - * ioctl. It is therefore necessary to hard-code the value for the ioctl - * using the size of the struct that is expected in any ABI other than - * PCuABI, to ensure that the value for MMC_IOC_CMD is unchanged. The - * original definition is as follows: - * - * #define MMC_IOC_CMD _IOWR(MMC_BLOCK_MAJOR, 0, struct mmc_ioc_cmd) - */ -#define MMC_IOC_CMD _IOC(_IOC_READ|_IOC_WRITE, MMC_BLOCK_MAJOR, 0, 72) +#define MMC_IOC_CMD _IOWR(MMC_BLOCK_MAJOR, 0, struct mmc_ioc_cmd) + /* * MMC_IOC_MULTI_CMD: Used to send an array of MMC commands described by * the structure mmc_ioc_multi_cmd. The MMC driver will issue all * commands in array in sequence to card. - * - * Note: the size of struct mmc_ioc_multi_cmd changes in PCuABI, due to the - * increased alignment of struct mmc_ioc_cmd. For that reason, its value needs - * to be hard-coded too. Original definition: - * - * #define MMC_IOC_MULTI_CMD _IOWR(MMC_BLOCK_MAJOR, 1, struct mmc_ioc_multi_cmd) */ -#define MMC_IOC_MULTI_CMD _IOC(_IOC_READ|_IOC_WRITE, MMC_BLOCK_MAJOR, 1, 8) +#define MMC_IOC_MULTI_CMD _IOWR(MMC_BLOCK_MAJOR, 1, struct mmc_ioc_multi_cmd) + /* * Since this ioctl is only meant to enhance (and not replace) normal access * to the mmc bus device, an upper data transfer limit of MMC_IOC_MAX_BYTES -- 2.34.1

5 months

2
1
0 0

[PATCH v5 0/3] drm purecap enablement

by carsten.haitzler＠foss.arm.com

From: Carsten Haitzler <carsten.haitzler(a)arm.com> This series starts to enable purecap support for drm ioctls. This series enables all the libdrm tests (tested and working). You will also need the purecap libdrm port as well to complement this. Carsten Haitzler (3): drm: Fix copy to/from user so that caps transport in the region drm: Fix purecap vblank handling drm: fix up purecap handling of all iotcls in libdrm test tools drivers/gpu/drm/drm_atomic_uapi.c | 98 +++++++++++++++++------ drivers/gpu/drm/drm_connector.c | 108 +++++++++++++++++++------ drivers/gpu/drm/drm_file.c | 2 +- drivers/gpu/drm/drm_ioctl.c | 6 +- drivers/gpu/drm/drm_mode_config.c | 126 +++++++++++++++++++++--------- drivers/gpu/drm/drm_mode_object.c | 37 ++++++++- drivers/gpu/drm/drm_plane.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_property.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_vblank.c | 39 ++++++--- include/drm/drm_vblank.h | 23 ++++++ include/uapi/drm/drm.h | 6 +- include/uapi/drm/drm_mode.h | 42 +++++----- tools/include/uapi/drm/drm.h | 2 +- 13 files changed, 506 insertions(+), 161 deletions(-) -- 2.25.1

5 months

1
3
0 0

[PATCH v4 0/3] drm purecap enablement

by carsten.haitzler＠foss.arm.com

From: Carsten Haitzler <carsten.haitzler(a)arm.com> This series starts to enable purecap support for drm ioctls. This series enables all the libdrm tests (tested and working). You will also need the purecap libdrm port as well to complement this. Carsten Haitzler (3): drm: Fix copy to/from user so that caps transport in the region drm: Fix purecap vblank handling drm: fix up purecap handling of all iotcls in libdrm test tools drivers/gpu/drm/drm_atomic_uapi.c | 98 +++++++++++++++++------ drivers/gpu/drm/drm_connector.c | 108 +++++++++++++++++++------ drivers/gpu/drm/drm_file.c | 2 +- drivers/gpu/drm/drm_ioctl.c | 6 +- drivers/gpu/drm/drm_mode_config.c | 126 +++++++++++++++++++++--------- drivers/gpu/drm/drm_mode_object.c | 37 ++++++++- drivers/gpu/drm/drm_plane.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_property.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_vblank.c | 39 ++++++--- include/drm/drm_vblank.h | 23 ++++++ include/uapi/drm/drm.h | 6 +- include/uapi/drm/drm_mode.h | 42 +++++----- tools/include/uapi/drm/drm.h | 2 +- 13 files changed, 506 insertions(+), 161 deletions(-) -- 2.25.1

5 months

3
8
0 0

[PATCH v3 0/3] drm purecap enablement

by carsten.haitzler＠foss.arm.com

From: Carsten Haitzler <carsten.haitzler(a)arm.com> This series starts to enable purecap support for drm ioctls. This series enables all the libdrm tests (tested and working). You will also need the purecap libdrm port as well to complement this. Carsten Haitzler (3): drm: Fix copy to/from user so that caps transport in the region drm: Fix purecap vblank handling drm: fix up purecap handling of all iotcls in libdrm test tools drivers/gpu/drm/drm_atomic_uapi.c | 98 +++++++++++++++++------ drivers/gpu/drm/drm_connector.c | 108 +++++++++++++++++++------ drivers/gpu/drm/drm_file.c | 2 +- drivers/gpu/drm/drm_ioctl.c | 6 +- drivers/gpu/drm/drm_mode_config.c | 126 +++++++++++++++++++++--------- drivers/gpu/drm/drm_mode_object.c | 37 ++++++++- drivers/gpu/drm/drm_plane.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_property.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_vblank.c | 39 ++++++--- include/drm/drm_vblank.h | 23 ++++++ include/uapi/drm/drm.h | 6 +- include/uapi/drm/drm_mode.h | 42 +++++----- tools/include/uapi/drm/drm.h | 6 +- 13 files changed, 508 insertions(+), 163 deletions(-) -- 2.25.1

5 months, 1 week

2
4
0 0

[PATCH v2] mmc: ioctl: Remove hard-coded MMC_IOC_{MULTI_}CMD values

by Akram Ahmad

The "mmc: Support capabilities in MMC_IOC_{MULTI_}CMD ioctls" patch previously hard-coded the ioctl command value in order to maintain compatibility with non-PCuABI ioctl handlers such as those found in block.c. This deviates from the standard approach which is to always define the ioctl command value in terms of the struct size (which may change), instead providing any fixes in the kernel via compat ioctl handlers. This patch therefore aligns the MMC ioctl handlers in block.c with the standard approach. Undoing the hard-coding in the header leaves the ioctl command value unchanged in all situations except in PCuABI where the structs sizes, and hence the ioctl command values, are now larger. (The ioctl value remains unchanged in compat64 as well). Signed-off-by: Akram Ahmad <Akram.Ahmad(a)arm.com> --- drivers/mmc/core/block.c | 62 ++++++++++++++++++++++++++++++++-- include/uapi/linux/mmc/ioctl.h | 21 +++++------- 2 files changed, 69 insertions(+), 14 deletions(-) diff --git a/drivers/mmc/core/block.c b/drivers/mmc/core/block.c index 17a1d24bdaed..d9b50bbd54ee 100644 --- a/drivers/mmc/core/block.c +++ b/drivers/mmc/core/block.c @@ -76,6 +76,15 @@ MODULE_ALIAS("mmc:block"); #define MMC_EXTRACT_INDEX_FROM_ARG(x) ((x & 0x00FF0000) >> 16) #define MMC_EXTRACT_VALUE_FROM_ARG(x) ((x & 0x0000FF00) >> 8) +/* + * The compat_mmc_ioc_{multi_}cmd structs have different sizes in PCuABI and in + * compat64, which affects the expected MMC_IOC_{MULTI_}CMD ioctl values. The + * following ioctl values reflect the sizes of the compat64 structs defined in + * this file. + */ +#define MMC_IOC_CMD_COMPAT _IOWR(MMC_BLOCK_MAJOR, 0, struct compat_mmc_ioc_cmd) +#define MMC_IOC_MULTI_CMD_COMPAT _IOWR(MMC_BLOCK_MAJOR, 1, struct compat_mmc_ioc_multi_cmd) + static DEFINE_MUTEX(block_mutex); /* @@ -891,7 +900,37 @@ static int mmc_blk_ioctl(struct block_device *bdev, blk_mode_t mode, static int mmc_blk_compat_ioctl(struct block_device *bdev, blk_mode_t mode, unsigned int cmd, unsigned long arg) { - return mmc_blk_ioctl(bdev, mode, cmd, (user_uintptr_t) compat_ptr(arg)); + struct mmc_blk_data *md; + int ret; + + switch (cmd) { + case MMC_IOC_CMD_COMPAT: + ret = mmc_blk_check_blkdev(bdev); + if (ret) + return ret; + md = mmc_blk_get(bdev->bd_disk); + if (!md) + return -EINVAL; + ret = mmc_blk_ioctl_cmd(md, + (struct mmc_ioc_cmd __user *)arg, + NULL); + mmc_blk_put(md); + return ret; + case MMC_IOC_MULTI_CMD_COMPAT: + ret = mmc_blk_check_blkdev(bdev); + if (ret) + return ret; + md = mmc_blk_get(bdev->bd_disk); + if (!md) + return -EINVAL; + ret = mmc_blk_ioctl_multi_cmd(md, + (struct mmc_ioc_multi_cmd __user *)arg, + NULL); + mmc_blk_put(md); + return ret; + default: + return -EINVAL; + } } #endif @@ -2700,7 +2739,26 @@ static long mmc_rpmb_ioctl(struct file *filp, unsigned int cmd, static long mmc_rpmb_ioctl_compat(struct file *filp, unsigned int cmd, unsigned long arg) { - return mmc_rpmb_ioctl(filp, cmd, (user_uintptr_t)compat_ptr(arg)); + struct mmc_rpmb_data *rpmb = filp->private_data; + int ret; + + switch (cmd) { + case MMC_IOC_CMD_COMPAT: + ret = mmc_blk_ioctl_cmd(rpmb->md, + (struct mmc_ioc_cmd __user *)arg, + rpmb); + break; + case MMC_IOC_MULTI_CMD_COMPAT: + ret = mmc_blk_ioctl_multi_cmd(rpmb->md, + (struct mmc_ioc_multi_cmd __user *)arg, + rpmb); + break; + default: + ret = -EINVAL; + break; + } + + return ret; } #endif diff --git a/include/uapi/linux/mmc/ioctl.h b/include/uapi/linux/mmc/ioctl.h index 9d34bfc49198..e99d932268c6 100644 --- a/include/uapi/linux/mmc/ioctl.h +++ b/include/uapi/linux/mmc/ioctl.h @@ -67,26 +67,23 @@ struct mmc_ioc_multi_cmd { /* * The size of struct mmc_ioc_cmd changes in PCuABI due to the use of * __kernel_uintptr_t, which in turn modifies the command value for this - * ioctl. It is therefore necessary to hard-code the value for the ioctl - * using the size of the struct that is expected in any ABI other than - * PCuABI, to ensure that the value for MMC_IOC_CMD is unchanged. The - * original definition is as follows: - * - * #define MMC_IOC_CMD _IOWR(MMC_BLOCK_MAJOR, 0, struct mmc_ioc_cmd) + * ioctl. As this command value is not hard-coded to represent the original + * size of the struct, any ioctl handlers in compat64 must handle this change + * in values instead. */ -#define MMC_IOC_CMD _IOC(_IOC_READ|_IOC_WRITE, MMC_BLOCK_MAJOR, 0, 72) +#define MMC_IOC_CMD _IOWR(MMC_BLOCK_MAJOR, 0, struct mmc_ioc_cmd) + /* * MMC_IOC_MULTI_CMD: Used to send an array of MMC commands described by * the structure mmc_ioc_multi_cmd. The MMC driver will issue all * commands in array in sequence to card. * * Note: the size of struct mmc_ioc_multi_cmd changes in PCuABI, due to the - * increased alignment of struct mmc_ioc_cmd. For that reason, its value needs - * to be hard-coded too. Original definition: - * - * #define MMC_IOC_MULTI_CMD _IOWR(MMC_BLOCK_MAJOR, 1, struct mmc_ioc_multi_cmd) + * increased alignment of struct mmc_ioc_cmd. Similarly to MMC_IOC_CMD, any + * ioctl handlers must account for this change when running in compat64. */ -#define MMC_IOC_MULTI_CMD _IOC(_IOC_READ|_IOC_WRITE, MMC_BLOCK_MAJOR, 1, 8) +#define MMC_IOC_MULTI_CMD _IOWR(MMC_BLOCK_MAJOR, 1, struct mmc_ioc_multi_cmd) + /* * Since this ioctl is only meant to enhance (and not replace) normal access * to the mmc bus device, an upper data transfer limit of MMC_IOC_MAX_BYTES -- 2.34.1

5 months, 1 week

2
3
0 0

[PATCH v2 0/3] drm purecap enablement

by carsten.haitzler＠foss.arm.com

From: Carsten Haitzler <carsten.haitzler(a)arm.com> This series starts to enable purecap support for drm ioctls. This series enables all the libdrm tests (tested and working). You will also need the purecap libdrm port as well to complement this. Carsten Haitzler (3): drm: Fix copy to/from user so that caps trasnport in the region drm: Fix purecap vblank handling drm: fix up purecap handling of all iotcls in libdrm test tools drivers/gpu/drm/drm_atomic_uapi.c | 3 +- drivers/gpu/drm/drm_connector.c | 108 +++++++++++++++++++------ drivers/gpu/drm/drm_file.c | 2 +- drivers/gpu/drm/drm_ioc32.c | 1 + drivers/gpu/drm/drm_ioctl.c | 8 +- drivers/gpu/drm/drm_mode_config.c | 126 +++++++++++++++++++++--------- drivers/gpu/drm/drm_mode_object.c | 37 ++++++++- drivers/gpu/drm/drm_plane.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_property.c | 89 +++++++++++++++++---- drivers/gpu/drm/drm_vblank.c | 39 ++++++--- include/drm/drm_vblank.h | 23 ++++++ include/uapi/drm/drm.h | 6 +- include/uapi/drm/drm_mode.h | 30 +++---- tools/include/uapi/drm/drm.h | 6 +- 14 files changed, 430 insertions(+), 137 deletions(-) -- 2.25.1

5 months, 1 week

3
12
0 0

[PATCH 0/3] drm purecap enablement

by carsten.haitzler＠foss.arm.com

From: Carsten Haitzler <carsten.haitzler(a)arm.com> This series starts to enable purecap support for drm ioctls. This series enables all the libdrm tests (tested and working). You will also need the purecap libdrm port as well to complement this. Carsten Haitzler (3): drm: Fix copy to/from user so that caps trasnport in the region drm: Fix purecap vblank handling drm: fix up purecap handling of all iotcls in libdrm test tools drivers/gpu/drm/drm_atomic_uapi.c | 3 +- drivers/gpu/drm/drm_connector.c | 107 +++++++++++++++++++------ drivers/gpu/drm/drm_file.c | 2 +- drivers/gpu/drm/drm_internal.h | 4 + drivers/gpu/drm/drm_ioc32.c | 8 +- drivers/gpu/drm/drm_ioctl.c | 20 ++--- drivers/gpu/drm/drm_mode_config.c | 126 +++++++++++++++++++++--------- drivers/gpu/drm/drm_mode_object.c | 37 ++++++++- drivers/gpu/drm/drm_plane.c | 88 +++++++++++++++++---- drivers/gpu/drm/drm_property.c | 87 ++++++++++++++++++--- drivers/gpu/drm/drm_vblank.c | 46 ++++++++--- include/drm/drm_drv.h | 9 +++ include/drm/drm_vblank.h | 23 ++++++ include/uapi/drm/drm.h | 6 +- include/uapi/drm/drm_mode.h | 34 ++++---- tools/include/uapi/drm/drm.h | 4 +- 16 files changed, 458 insertions(+), 146 deletions(-) -- 2.25.1

5 months, 1 week

3
24
0 0

[PATCH 00/12] KVM: arm64: morello: Basic support

by Beata Michalska

Hi All, This series adds bare minimum required to support kvm virtual machines. The majority of changes make sure that executing Morello instructions by the hypervisor/guest is safe and that core capability registers are properly set and maintained across context switches. Additionally, support for COMPAT64 allows plain aarch64 tools to still manage virtual machines. Support for purecap guests is yet to come. The changes have been tested with KVM kselftests (in plain aarch64) in their default setup (some of the tests do expose number of options and those have not been fully exercised). The changes are available at: https://git.morello-project.org/Bea/linux/-/tree/morello/kvm_base_2?ref_typ… Beata Michalska (12): KVM: arm64: Provide explicit capability annotations KVM: arm64: morello: Add (V)TCR_EL2 HW field descriptors KVM: arm64: morello: Enable page-based hw attributes arm64/sysreg: morello: Support for capability registers arm64/sysreg: morello: Add capability-aware sysreg helpers KVM: arm64: morello: Add sysreg access capability variants KVM: arm64: morello: Extend the context by capability sysregs KVM: arm64: morello: Provide valid capability on return from exception KVM: arm64: morello: Disable trapping of Morello instructions KVM: arm64: morello: Preserve capabilities on context save / restore KVM: arm64: morello: Enable support for COMPAT64 KVM: arm64: morello: Enable KVM Documentation/virt/kvm/api.rst | 10 +- arch/arm64/include/asm/kvm_arm.h | 12 ++ arch/arm64/include/asm/kvm_asm.h | 108 ++++++++++- arch/arm64/include/asm/kvm_emulate.h | 5 + arch/arm64/include/asm/kvm_host.h | 38 ++++ arch/arm64/include/asm/kvm_hyp.h | 74 ++++++-- arch/arm64/include/asm/kvm_pgtable.h | 20 ++ arch/arm64/include/asm/sysreg.h | 83 +++++++++ arch/arm64/include/uapi/asm/kvm.h | 6 +- arch/arm64/kernel/asm-offsets.c | 3 + arch/arm64/kernel/hyp-stub.S | 5 + arch/arm64/kvm/Kconfig | 1 - arch/arm64/kvm/arch_timer.c | 4 +- arch/arm64/kvm/arm.c | 207 ++++++++++++++------- arch/arm64/kvm/guest.c | 10 +- arch/arm64/kvm/hyp/entry.S | 12 +- arch/arm64/kvm/hyp/include/hyp/switch.h | 4 +- arch/arm64/kvm/hyp/include/hyp/sysreg-sr.h | 58 +++++- arch/arm64/kvm/hyp/nvhe/host.S | 66 ++++++- arch/arm64/kvm/hyp/nvhe/hyp-init.S | 14 ++ arch/arm64/kvm/hyp/nvhe/switch.c | 9 +- arch/arm64/kvm/hyp/nvhe/sys_regs.c | 4 +- arch/arm64/kvm/hyp/pgtable.c | 10 +- arch/arm64/kvm/hyp/vhe/switch.c | 11 +- arch/arm64/kvm/hypercalls.c | 4 +- arch/arm64/kvm/pmu-emul.c | 8 +- arch/arm64/kvm/sys_regs.c | 21 ++- arch/arm64/kvm/vgic/vgic-its.c | 8 +- arch/arm64/kvm/vgic/vgic-kvm-device.c | 11 +- include/linux/kvm_host.h | 8 +- include/uapi/linux/kvm.h | 10 +- tools/arch/arm64/include/uapi/asm/kvm.h | 6 +- tools/include/uapi/linux/kvm.h | 12 +- virt/kvm/Kconfig | 2 +- virt/kvm/kvm_main.c | 99 +++++++--- virt/kvm/vfio.c | 2 +- 36 files changed, 776 insertions(+), 189 deletions(-) -- 2.25.1

5 months, 1 week

2
32
0 0

[PATCH] arm64: compat: Create null-derived out-of-bounds pointers

by Kevin Brodsky

In PCuABI, we need compat_ptr() to create a valid capability from an arbitrary compat pointer, to enable uaccess to the underlying memory. However, if the compat pointer is clearly an invalid address, it is preferable to return a null-derived (invalid) capability. This is especially true when the address is null: we clearly want to return the null capability in that case. This should be a non-functional change (uaccess would fail anyway due to the absence of underlying mapping), but it avoids some confusion when debugging compat64 handling (and reduces the creation of capabilities). Signed-off-by: Kevin Brodsky <kevin.brodsky(a)arm.com> --- Review branch: https://git.morello-project.org/kbrodsky-arm/linux/-/commits/morello/null_c… arch/arm64/include/asm/compat.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/arch/arm64/include/asm/compat.h b/arch/arm64/include/asm/compat.h index 4556bed1704f..acf5487985b5 100644 --- a/arch/arm64/include/asm/compat.h +++ b/arch/arm64/include/asm/compat.h @@ -93,7 +93,9 @@ static inline void __user *compat_ptr(compat_uptr_t uptr) * TODO [Morello] - this should be done using the current user DDC, not * the root user capability. */ - return (void __user *)cheri_address_set(cheri_user_root_allperms_cap, uptr); + return likely(uptr >= PAGE_SIZE && uptr < TASK_SIZE_MAX) ? + (void __user *)cheri_address_set(cheri_user_root_allperms_cap, uptr) : + as_user_ptr(uptr); } #define compat_ptr(uptr) compat_ptr(uptr) #endif -- 2.43.0

5 months, 1 week

2
3
0 0

2024

2023

2022

linux-morello