Changes in os.h

33c9da5d - um: Rewrite the sigio workaround based on epoll and tgkill

Sat, 15 Mar 2025 16:19:08 +0000

um: Rewrite the sigio workaround based on epoll and tgkillThe existing sigio workaround implementation removes FDs from thepoll when events are triggered, requiring users to re-add them viaadd_sigio_fd() after processing. This introduces a potential racecondition between FD removal in write_sigio_thread() and next_pollupdate in __add_sigio_fd(), and is inefficient due to frequent FDremoval and re-addition. Rewrite the implementation based on epolland tgkill for improved efficiency and reliability.Signed-off-by: Tiwei Bie Link: https://patch.msgid.link/20250315161910.4082396-2-tiwei.btw@antgroup.comSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

4f087eaf - um: Add pthread-based helper support

Wed, 19 Mar 2025 13:55:20 +0000

um: Add pthread-based helper supportIntroduce a new set of utility functions that can be used to createpthread-based helpers. Helper threads created in this way will ensurethread safety for errno while sharing the same memory space.Signed-off-by: Tiwei Bie Link: https://patch.msgid.link/20250319135523.97050-2-tiwei.btw@antgroup.comSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

84a6fc37 - um: remove copy_from_kernel_nofault_allowed

Mon, 10 Feb 2025 16:09:26 +0000

um: remove copy_from_kernel_nofault_allowedThere is no need to override the default version of this functionanymore as UML now has proper _nofault memory access functions.Doing this also fixes the fact that the implementation was incorrect asusing mincore() will incorrectly flag pages as inaccessible if they wereswapped out by the host.Fixes: f75b1b1bedfb ("um: Implement probe_kernel_read()")Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20250210160926.420133-3-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

ce6e85a1 - um: remove broken double fault detection

Sun, 03 Nov 2024 15:05:06 +0000

um: remove broken double fault detectionThe show_stack function had some code to detect double faults. However,the logic is wrong and it would e.g. trigger if a WARNING happenedinside an IRQ.Remove it without trying to add a new logic. The current behaviour,which will just fault repeatedly until the IRQ stack is used up and thehost kills UML, seems to be good enough.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20241103150506.1367695-5-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

37c69115 - um: remove file sync for stub data

Sun, 03 Nov 2024 15:05:04 +0000

um: remove file sync for stub dataThere is no need to sync the stub code to "disk" for the other processto see the correct memory. Drop the fsync there and remove the helperfunction.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20241103150506.1367695-3-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

4e5adbe4 - um: Add os_set_pdeathsig helper function

Thu, 24 Oct 2024 14:28:25 +0000

um: Add os_set_pdeathsig helper functionThis helper can be used to set the parent-death signal of the callingprocess to SIGKILL to ensure that the process will be killed if theUML kernel dies unexpectedly without proper cleanup. This helper willbe used in the follow-up patches.Signed-off-by: Tiwei Bie Link: https://patch.msgid.link/20241024142828.2612828-2-tiwei.btw@antgroup.comSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

2717c6b6 - um: Abandon the _PAGE_NEWPROT bit

Fri, 11 Oct 2024 10:23:53 +0000

um: Abandon the _PAGE_NEWPROT bitWhen a PTE is updated in the page table, the _PAGE_NEWPAGE bit willalways be set. And the corresponding page will always be mapped orunmapped depending on whether the PTE is present or not. The checkon the _PAGE_NEWPROT bit is not really reachable. Abandoning it willallow us to simplify the code and remove the unreachable code.Reviewed-by: Benjamin Berg Signed-off-by: Tiwei Bie Link: https://patch.msgid.link/20241011102354.1682626-2-tiwei.btw@antgroup.comSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

68b9883c - um: Discover host_task_size from envp

Thu, 19 Sep 2024 12:45:09 +0000

um: Discover host_task_size from envpWhen loading the UML binary, the host kernel will place the stack at thehighest possible address. It will then map the program name andenvironment variables onto the start of the stack.As such, an easy way to figure out the host_task_size is to use thehighest pointer to an environment variable as a reference.Ensure that this works by disabling address layout randomization andre-executing UML in case it was enabled.This increases the available TASK_SIZE for 64 bit UML considerably.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240919124511.282088-9-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

c6ce7200 - um: remove auxiliary FP registers

Fri, 04 Oct 2024 23:38:21 +0000

um: remove auxiliary FP registersWe do not need the extra save/restore of the FP registers when gettingthe fault information. This was originally added in commit 2f56debd77a8("uml: fix FP register corruption") but at that time the code was notsaving/restoring the FP registers when switching to userspace. This wasfixed in commit fbfe9c847edf ("um: Save FPU registers between taskswitches") and since then the auxiliary registers have not been useful.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20241004233821.2130874-1-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

71fae9df - um: Remove unused os_getpgrp function

Fri, 13 Sep 2024 13:44:41 +0000

um: Remove unused os_getpgrp functionThe function is not used anywhere.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240913134442.967599-5-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

377c23c5 - um: Remove unused os_stop_process

Fri, 13 Sep 2024 13:44:40 +0000

um: Remove unused os_stop_processThe function is not used anywhere.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240913134442.967599-4-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

47e17496 - um: Remove unused os_process_parent

Fri, 13 Sep 2024 13:44:39 +0000

um: Remove unused os_process_parentThe function is not used anywhere.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240913134442.967599-3-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

7852ee06 - um: Remove unused os_process_pc

Fri, 13 Sep 2024 13:44:38 +0000

um: Remove unused os_process_pcThe function is not used anywhere in the codebase.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240913134442.967599-2-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

573a446f - um: simplify and consolidate TLB updates

Wed, 03 Jul 2024 13:45:35 +0000

um: simplify and consolidate TLB updatesThe HVC update was mostly used to compress consecutive calls into one.This is mostly relevant for userspace where it is already handled by thesyscall stub code.Simplify the whole logic and consolidate it for both kernel anduserspace. This does remove the sequential syscall compression for thekernel, however that shouldn't be the main factor in most runs.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240703134536.1161108-12-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

3c83170d - um: Delay flushing syscalls until the thread is restarted

Wed, 03 Jul 2024 13:45:32 +0000

um: Delay flushing syscalls until the thread is restartedAs running the syscalls is expensive due to context switches, we shoulddo so as late as possible in case more syscalls need to be queued lateron. This will also benefit a later move to a SECCOMP enabled userspaceas in that case the need for extra context switches is removed entirely.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240703134536.1161108-9-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

a5d2cfe7 - um: remove copy_context_skas0

Wed, 03 Jul 2024 13:45:31 +0000

um: remove copy_context_skas0The kernel flushes the memory ranges anyway for CoW and does not assumethat the userspace process has anything set up already. So, start with afresh process for the new mm context.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240703134536.1161108-8-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

76ed9158 - um: Rework syscall handling

Wed, 03 Jul 2024 13:45:28 +0000

um: Rework syscall handlingRework syscall handling to be platform independent. Also create a cleansplit between queueing of syscalls and flushing them out, removing theneed to keep state in the code that triggers the syscalls.The code adds syscall_data_len to the global mm_id structure. This willbe used later to allow surrounding code to track whether syscalls stillneed to run and if errors occurred.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20240703134536.1161108-5-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

e20f9b3c - um: add mmap/mremap OS calls

Tue, 02 Jul 2024 17:21:20 +0000

um: add mmap/mremap OS callsFor the upcoming shared-memory time-travel externaloptimisations, we need to be able to mmap/mremap.Add the necessary OS calls.Link: https://patch.msgid.link/20240702192118.ca4472963638.Ic2da1d3a983fe57340c1b693badfa9c5bd2d8c61@changeidSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

5cde6096 - um: generalize os_rcv_fd

Tue, 02 Jul 2024 17:21:19 +0000

um: generalize os_rcv_fdChange os_rcv_fd() to os_rcv_fd_msg() that can more generallyreceive any number of FDs in any kind of message.Link: https://patch.msgid.link/20240702192118.40b78b2bfe4e.Ic6ec12d72630e5bcae1e597d6bd5c6f29f441563@changeidSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h

b2f9b77c - um: chan: use blocking IO for console output for time-travel

Wed, 18 Oct 2023 12:36:43 +0000

um: chan: use blocking IO for console output for time-travelWhen in time-travel mode (infinite-cpu or external) time should not passfor writing to the console. As such, it makes sense to put the FD forthe output side into blocking mode and simply let any write to it hang.If we did not do this, then time could pass waiting for the console tobecome writable again. This is not desirable as it has random effects onthe clock between runs.Implement this by duplicating the FD if output is active in a relevantmode and setting the duplicate to be blocking. This avoids changing theinput channel to be blocking should it exists. After this, use theblocking FD for all write operations and do not allocate an IRQ it isset.Without time-travel mode fd_out will always match fd_in and IRQs areregistered.Signed-off-by: Benjamin Berg Link: https://patch.msgid.link/20231018123643.1255813-4-benjamin@sipsolutions.netSigned-off-by: Johannes Berg List of files: /linux-6.15/arch/um/include/shared/os.h