PCI/sysfs: Add lockdown checks to legacy I/O and memory handlers
PATCH 1/1 PCI/sysfs: Add lockdown checks to legacy I/O and memory handlers
Currently, the legacy I/O and memory sysfs handlers do not check security_locked_down(LOCKDOWN_PCI_ACCESS), leaving the legacy_io and legacy_mem files unprotected when the kernel is locked down.
Commit eb627e17727e ("PCI: Lock down BAR access when the kernel is locked down") added the check to pci_write_config(), pci_mmap_resource(), and pci_write_resource_io() to prevent userspace from programming DMA-capable hardware that could be used to modify kernel code, but did not cover the legacy handlers.
As a result, root can still write arbitrary I/O ports and map the legacy I/O and memory spaces while the kernel is locked down, which is the same capability the lockdown is meant to remove.
Thus, add the same check to pci_write_legacy_io(), pci_mmap_legacy_mem(), and pci_mmap_legacy_io().
These generic handlers cover both architectures that define HAVE_PCI_LEGACY (such as Alpha and PowerPC).
--- drivers/pci/pci-sysfs.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+)
drivers/pci/pci-sysfs.c +15 -0
@@ -913,6 +913,11 @@ static ssize_t pci_write_legacy_io(struct file *filp, struct kobject *kobj, char *buf, loff_t off, size_t count) {struct pci_bus *bus =to_pci_bus (kobj_to_dev (kobj)); +int ret; + + ret =security_locked_down (LOCKDOWN_PCI_ACCESS ); +if (ret) +return ret;/* Only support 1, 2 or 4 byte accesses */ if (count !=1 && count !=2 && count !=4 )
@@ -937,6 +942,11 @@ static int pci_mmap_legacy_mem(struct file *filp, struct kobject *kobj, struct vm_area_struct *vma) {struct pci_bus *bus =to_pci_bus (kobj_to_dev (kobj)); +int ret; + + ret =security_locked_down (LOCKDOWN_PCI_ACCESS ); +if (ret) +return ret;return pci_mmap_legacy_page_range (bus, vma, pci_mmap_mem); }
@@ -957,6 +967,11 @@ static int pci_mmap_legacy_io(struct file *filp, struct kobject *kobj, struct vm_area_struct *vma) {struct pci_bus *bus =to_pci_bus (kobj_to_dev (kobj)); +int ret; + + ret =security_locked_down (LOCKDOWN_PCI_ACCESS ); +if (ret) +return ret;return pci_mmap_legacy_page_range (bus, vma, pci_mmap_io); }
Signature · 1 line
2.55.0
[email protected]
AI reviewSashiko has reviewed this patch and found no issues. It looks great!
Quoted 17 lines
Currently, the legacy I/O and memory sysfs handlers do not check
security_locked_down(LOCKDOWN_PCI_ACCESS), leaving thelegacy_ioandlegacy_memfiles unprotected when the kernel is locked down.Commit
eb627e17727e("PCI: Lock down BAR access when the kernel is locked down") added the check topci_write_config(),pci_mmap_resource(), andpci_write_resource_io()to prevent userspace from programming DMA-capable hardware that could be used to modify kernel code, but did not cover the legacy handlers.As a result, root can still write arbitrary I/O ports and map the legacy I/O and memory spaces while the kernel is locked down, which is the same capability the lockdown is meant to remove.
Quoted 8 lines
Thus, add the same check to
pci_write_legacy_io(),pci_mmap_legacy_mem(), andpci_mmap_legacy_io().These generic handlers cover both architectures that define
HAVE_PCI_LEGACY(such as Alpha and PowerPC).Fixes eb627e17727e ("PCI: Lock down BAR access when the kernel is locked down")Signed-off-by Krzysztof Wilczyński <[email protected]>
Sashiko has reviewed this patch and found no issues. It looks great!
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Krzysztof Wilczyński
AppliedHello,
Quoted 17 lines
Currently, the legacy I/O and memory sysfs handlers do not check
security_locked_down(LOCKDOWN_PCI_ACCESS), leaving thelegacy_ioandlegacy_memfiles unprotected when the kernel is locked down.Commit
eb627e17727e("PCI: Lock down BAR access when the kernel is locked down") added the check topci_write_config(),pci_mmap_resource(), andpci_write_resource_io()to prevent userspace from programming DMA-capable hardware that could be used to modify kernel code, but did not cover the legacy handlers.As a result, root can still write arbitrary I/O ports and map the legacy I/O and memory spaces while the kernel is locked down, which is the same capability the lockdown is meant to remove.
Quoted 5 lines
Thus, add the same check to
pci_write_legacy_io(),pci_mmap_legacy_mem(), andpci_mmap_legacy_io().These generic handlers cover both architectures that define
HAVE_PCI_LEGACY(such as Alpha and PowerPC).
Applied to the sysfs branch.
Thank you!
Krzysztof
Hi,
Quoted 35 lines · Krzysztof Wilczyński
Currently, the legacy I/O and memory sysfs handlers do not check
security_locked_down(LOCKDOWN_PCI_ACCESS), leaving thelegacy_ioandlegacy_memfiles unprotected when the kernel is locked down.Commit
eb627e17727e("PCI: Lock down BAR access when the kernel is locked down") added the check topci_write_config(),pci_mmap_resource(), andpci_write_resource_io()to prevent userspace from programming DMA-capable hardware that could be used to modify kernel code, but did not cover the legacy handlers.As a result, root can still write arbitrary I/O ports and map the legacy I/O and memory spaces while the kernel is locked down, which is the same capability the lockdown is meant to remove.
Thus, add the same check to
pci_write_legacy_io(),pci_mmap_legacy_mem(), andpci_mmap_legacy_io().These generic handlers cover both architectures that define
HAVE_PCI_LEGACY(such as Alpha and PowerPC).Fixes eb627e17727e ("PCI: Lock down BAR access when the kernel is locked down")Signed-off-by Krzysztof Wilczyński <[email protected]>--- drivers/pci/pci-sysfs.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+)drivers/pci/pci-sysfs.c +0
-0@@ -913,6 +913,11 @@ static ssize_t pci_write_legacy_io(struct file *filp, struct kobject *kobj, char *buf, loff_t off, size_t count)
drivers/pci/pci-sysfs.c static ssize_t pci_write_legacy_io(struct file *filp, struct kobject *kobj, · line 923
Quoted 10 lines
drivers/pci/pci-sysfs.c +5
-0{struct pci_bus *bus =to_pci_bus (kobj_to_dev (kobj)); +int ret; + + ret =security_locked_down (LOCKDOWN_PCI_ACCESS ); +if (ret) +return ret;/* Only support 1, 2 or 4 byte accesses */ if (count !=1 && count !=2 && count !=4 )
This hunk was misapplied AFAICT. The commit 747b9bbbbdfd ("PCI/sysfs: Add lockdown checks to legacy I/O and memory handlers") applies the hunk to __pci_legacy_is_visible() instead.
Two effects:
1) pci_write_legacy_io() is still not covered,
2) with lockdown, __pci_legacy_is_visible() now returns EPERM which should trigger a WARN() in create_files() and it would create the file with 0664 (!), if I am looking correctly (and Claude confirms).
Quoted 24 lines
drivers/pci/pci-sysfs.c +10
-0@@ -937,6 +942,11 @@ static int pci_mmap_legacy_mem(struct file *filp, struct kobject *kobj, struct vm_area_struct *vma) {struct pci_bus *bus =to_pci_bus (kobj_to_dev (kobj)); +int ret; + + ret =security_locked_down (LOCKDOWN_PCI_ACCESS ); +if (ret) +return ret;return pci_mmap_legacy_page_range (bus, vma, pci_mmap_mem); }@@ -957,6 +967,11 @@ static int pci_mmap_legacy_io(struct file *filp, struct kobject *kobj, struct vm_area_struct *vma) {struct pci_bus *bus =to_pci_bus (kobj_to_dev (kobj)); +int ret; + + ret =security_locked_down (LOCKDOWN_PCI_ACCESS ); +if (ret) +return ret;return pci_mmap_legacy_page_range (bus, vma, pci_mmap_io); }
Signature · 2 lines
js suse labs
Hello,
Quoted 11 lines
Quoted 11 lines
drivers/pci/pci-sysfs.c +3
-0@@ -913,6 +913,11 @@ static ssize_t pci_write_legacy_io(struct file *filp, struct kobject *kobj, char *buf, loff_t off, size_t count) {struct pci_bus *bus =to_pci_bus (kobj_to_dev (kobj)); +int ret; + + ret =security_locked_down (LOCKDOWN_PCI_ACCESS );
Quoted 8 lines
drivers/pci/pci-sysfs.c static ssize_t pci_write_legacy_io(struct file *filp, struct kobject *kobj, · line 923
Quoted 4 lines
drivers/pci/pci-sysfs.c +2
-0+if (ret) +return ret;/* Only support 1, 2 or 4 byte accesses */ if (count !=1 && count !=2 && count !=4 )This hunk was misapplied AFAICT. The commit
747b9bbbbdfd("PCI/sysfs: Add lockdown checks to legacy I/O and memory handlers") applies the hunk to__pci_legacy_is_visible()instead.
Yes, sadly it was misapplied. We found out recently, too. I will sent a patch to correct this a later today.
Thank you for letting us know, Jiri. Appreciated.
All the best,
Krzysztof