kernel/arch/x86/include/asm/segment.h, branch linux-rolling-stable

x86/vdso: Fix output operand size of RDPID

2025-10-15T10:03:17Z

[ Upstream commit ac9c408ed19d535289ca59200dd6a44a6a2d6036 ] RDPID instruction outputs to a word-sized register (64-bit on x86_64 and 32-bit on x86_32). Use an unsigned long variable to store the correct size. LSL outputs to 32-bit register, use %k operand prefix to always print the 32-bit name of the register. Use RDPID insn mnemonic while at it as the minimum binutils version of 2.30 supports it. [ bp: Merge two patches touching the same function into a single one. ] Fixes: ffebbaedc861 ("x86/vdso: Introduce helper functions for CPU and node number") Signed-off-by: Uros Bizjak Signed-off-by: Borislav Petkov (AMD) Link: https://lore.kernel.org/20250616095315.230620-1-ubizjak@gmail.com Signed-off-by: Sasha Levin

x86/headers: Replace ASSEMBLY with ASSEMBLER in non-UAPI headers

2025-03-19T10:47:30Z

While the GCC and Clang compilers already define __ASSEMBLER__ automatically when compiling assembly code, __ASSEMBLY__ is a macro that only gets defined by the Makefiles in the kernel. This can be very confusing when switching between userspace and kernelspace coding, or when dealing with UAPI headers that rather should use __ASSEMBLER__ instead. So let's standardize on the __ASSEMBLER__ macro that is provided by the compilers now. This is mostly a mechanical patch (done with a simple "sed -i" statement), with some manual tweaks in , and that mentioned this macro in comments with some missing underscores. Signed-off-by: Thomas Huth Signed-off-by: Ingo Molnar Cc: Brian Gerst Cc: Juergen Gross Cc: H. Peter Anvin Cc: Kees Cook Cc: Linus Torvalds Link: https://lore.kernel.org/r/20250314071013.1575167-38-thuth@redhat.com

x86/vdso: Choose the right GDT_ENTRY_CPUNODE for 32-bit getcpu() on 64-bit kernel

2023-08-08T07:31:43Z

The vDSO getcpu() reads CPU ID from the GDT_ENTRY_CPUNODE entry when the RDPID instruction is not available. And GDT_ENTRY_CPUNODE is defined as 28 on 32-bit Linux kernel and 15 on 64-bit. But the 32-bit getcpu() on 64-bit Linux kernel is compiled with 32-bit Linux kernel GDT_ENTRY_CPUNODE, i.e., 28, beyond the 64-bit Linux kernel GDT limit. Thus, it just fails _silently_. When BUILD_VDSO32_64 is defined, choose the 64-bit Linux kernel GDT definitions to compile the 32-bit getcpu(). Fixes: 877cff5296faa6e ("x86/vdso: Fake 32bit VDSO build on 64bit compile for vgetcpu") Reported-by: kernel test robot Reported-by: Shan Kang Signed-off-by: Xin Li Signed-off-by: Thomas Gleixner Link: https://lore.kernel.org/r/20230322061758.10639-1-xin3.li@intel.com Link: https://lore.kernel.org/oe-lkp/202303020903.b01fd1de-yujie.liu@intel.com

x86/cpu: Provide the full setup for getcpu() on x86-32

2023-02-06T14:48:54Z

setup_getcpu() configures two things: - it writes the current CPU & node information into MSR_TSC_AUX - it writes the same information as a GDT entry. By using the "full" setup_getcpu() on i386 it is possible to read the CPU information in userland via RDTSCP() or via LSL from the GDT. Provide an GDT_ENTRY_CPUNODE for x86-32 and make the setup function unconditionally available. Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Thomas Gleixner Reviewed-by: Roland Mainz Link: https://lore.kernel.org/r/20221125094216.3663444-2-bigeasy@linutronix.de

x86/signal/32: Merge native and compat 32-bit signal code

2022-10-19T07:58:49Z

There are significant differences between signal handling on 32-bit vs. 64-bit, like different structure layouts and legacy syscalls. Instead of duplicating that code for native and compat, merge both versions into one file. Signed-off-by: Brian Gerst Signed-off-by: Borislav Petkov Acked-by: "Eric W. Biederman" Link: https://lore.kernel.org/r/20220606203802.158958-8-brgerst@gmail.com Signed-off-by: Borislav Petkov

x86: Remove __USER32_DS

2022-10-19T07:58:49Z

Replace all users with the equivalent __USER_DS, which will make merging native and compat code simpler. Signed-off-by: Brian Gerst Signed-off-by: Borislav Petkov Acked-by: "Eric W. Biederman" Link: https://lore.kernel.org/r/20220606203802.158958-5-brgerst@gmail.com Signed-off-by: Borislav Petkov

x86/asm: Merge load_gs_index()

2022-04-14T12:15:54Z

Merge the 32- and 64-bit implementations of load_gs_index(). Signed-off-by: Brian Gerst Signed-off-by: Borislav Petkov Reviewed-by: Thomas Gleixner Acked-by: Andy Lutomirski Link: https://lore.kernel.org/r/20220325153953.162643-5-brgerst@gmail.com

x86/32: Remove lazy GS macros

2022-04-14T12:09:43Z

GS is always a user segment now. Signed-off-by: Brian Gerst Signed-off-by: Borislav Petkov Reviewed-by: Thomas Gleixner Acked-by: Andy Lutomirski Link: https://lore.kernel.org/r/20220325153953.162643-4-brgerst@gmail.com

x86/ibt,entry: Sprinkle ENDBR dust

2022-03-15T09:32:35Z

Kernel entry points should be having ENDBR on for IBT configs. The SYSCALL entry points are found through taking their respective address in order to program them in the MSRs, while the exception entry points are found through UNWIND_HINT_IRET_REGS. The rule is that any UNWIND_HINT_IRET_REGS at sym+0 should have an ENDBR, see the later objtool ibt validation patch. Signed-off-by: Peter Zijlstra (Intel) Acked-by: Josh Poimboeuf Link: https://lore.kernel.org/r/20220308154317.933157479@infradead.org

x86/ibt,xen: Sprinkle the ENDBR