- 08 Jul, 2005 40 commits
-
-
Jeff Dike authored
UML has had two modes of operation - an insecure, slow mode (tt mode) in which the kernel is mapped into every process address space which requires no host kernel modifications, and a secure, faster mode (skas mode) in which the UML kernel is in a separate host address space, which requires a patch to the host kernel. This patch implements something very close to skas mode for hosts which don't support skas - I'm calling this skas0. It provides the security of the skas host patch, and some of the performance gains. The two main things that are provided by the skas patch, /proc/mm and PTRACE_FAULTINFO, are implemented in a way that require no host patch. For the remote address space changing stuff (mmap, munmap, and mprotect), we set aside two pages in the process above its stack, one of which contains a little bit of code which can call mmap et al. To update the address space, the system call information (system call number and arguments) are written to the stub page above the code. The %esp is set to the beginning of the data, the %eip is set the the start of the stub, and it repeatedly pops the information into its registers and makes the system call until it sees a system call number of zero. This is to amortize the cost of the context switch across multiple address space updates. When the updates are done, it SIGSTOPs itself, and the kernel process continues what it was doing. For a PTRACE_FAULTINFO replacement, we set up a SIGSEGV handler in the child, and let it handle segfaults rather than nullifying them. The handler is in the same page as the mmap stub. The second page is used as the stack. The handler reads cr2 and err from the sigcontext, sticks them at the base of the stack in a faultinfo struct, and SIGSTOPs itself. The kernel then reads the faultinfo and handles the fault. A complication on x86_64 is that this involves resetting the registers to the segfault values when the process is inside the kill system call. This breaks on x86_64 because %rcx will contain %rip because you tell SYSRET where to return to by putting the value in %rcx. So, this corrupts $rcx on return from the segfault. To work around this, I added an arch_finish_segv, which on x86 does nothing, but which on x86_64 ptraces the child back through the sigreturn. This causes %rcx to be restored by sigreturn and avoids the corruption. Ultimately, I think I will replace this with the trick of having it send itself a blocked signal which will be unblocked by the sigreturn. This will allow it to be stopped just after the sigreturn, and PTRACE_SYSCALLed without all the back-and-forth of PTRACE_SYSCALLing it through sigreturn. This runs on a stock host, so theoretically (and hopefully), tt mode isn't needed any more. We need to make sure that this is better in every way than tt mode, though. I'm concerned about the speed of address space updates and page fault handling, since they involve extra round-trips to the child. We can amortize the round-trip cost for large address space updates by writing all of the operations to the data page and having the child execute them all at the same time. This will help fork and exec, but not page faults, since they involve only one page. I can't think of any way to help page faults, except to add something like PTRACE_FAULTINFO to the host. There is PTRACE_SIGINFO, but UML doesn't use siginfo for SIGSEGV (or anything else) because there isn't enough information in the siginfo struct to handle page faults (the faulting operation type is missing). Adding that would make PTRACE_SIGINFO a usable equivalent to PTRACE_FAULTINFO. As for the code itself: - The system call stub is in arch/um/kernel/sys-$(SUBARCH)/stub.S. It is put in its own section of the binary along with stub_segv_handler in arch/um/kernel/skas/process.c. This is manipulated with run_syscall_stub in arch/um/kernel/skas/mem_user.c. syscall_stub will execute any system call at all, but it's only used for mmap, munmap, and mprotect. - The x86_64 stub calls sigreturn by hand rather than allowing the normal sigreturn to happen, because the normal sigreturn is a SA_RESTORER in UML's address space provided by libc. Needless to say, this is not available in the child's address space. Also, it does a couple of odd pops before that which restore the stack to the state it was in at the time the signal handler was called. - There is a new field in the arch mmu_context, which is now a union. This is the pid to be manipulated rather than the /proc/mm file descriptor. Code which deals with this now checks proc_mm to see whether it should use the usual skas code or the new code. - userspace_tramp is now used to create a new host process for every UML process, rather than one per UML processor. It checks proc_mm and ptrace_faultinfo to decide whether to map in the pages above its stack. - start_userspace now makes CLONE_VM conditional on proc_mm since we need separate address spaces now. - switch_mm_skas now just sets userspace_pid[0] to the new pid rather than PTRACE_SWITCH_MM. There is an addition to userspace which updates its idea of the pid being manipulated each time around the loop. This is important on exec, when the pid will change underneath userspace(). - The stub page has a pte, but it can't be mapped in using tlb_flush because it is part of tlb_flush. This is why it's required for it to be mapped in by userspace_tramp. Other random things: - The stub section in uml.lds.S is page aligned. This page is written out to the backing vm file in setup_physmem because it is mapped from there into user processes. - There's some confusion with TASK_SIZE now that there are a couple of extra pages that the process can't use. TASK_SIZE is considered by the elf code to be the usable process memory, which is reasonable, so it is decreased by two pages. This confuses the definition of USER_PGDS_IN_LAST_PML4, making it too small because of the rounding down of the uneven division. So we round it to the nearest PGDIR_SIZE rather than the lower one. - I added a missing PT_SYSCALL_ARG6_OFFSET macro. - um_mmu.h was made into a userspace-usable file. - proc_mm and ptrace_faultinfo are globals which say whether the host supports these features. - There is a bad interaction between the mm.nr_ptes check at the end of exit_mmap, stack randomization, and skas0. exit_mmap will stop freeing pages at the PGDIR_SIZE boundary after the last vma. If the stack isn't on the last page table page, the last pte page won't be freed, as it should be since the stub ptes are there, and exit_mmap will BUG because there is an unfreed page. To get around this, TASK_SIZE is set to the next lowest PGDIR_SIZE boundary and mm->nr_ptes is decremented after the calls to init_stub_pte. This ensures that we know the process stack (and all other process mappings) will be below the top page table page, and thus we know that mm->nr_ptes will be one too many, and can be decremented. Things that need fixing: - We may need better assurrences that the stub code is PIC. - The stub pte is set up in init_new_context_skas. - alloc_pgdir is probably the right place. Signed-off-by:
Jeff Dike <jdike@addtoit.com> Signed-off-by:
Andrew Morton <akpm@osdl.org> Signed-off-by:
Linus Torvalds <torvalds@osdl.org>
-
Pavel Machek authored
freezeable() already tests for TRACED/STOPPED processes, no need to do it twice. Signed-off-by:
Pavel Machek <pavel@suse.cz> Signed-off-by:
-
Pavel Machek authored
Fix error handling and whitespace in swsusp.c. swsusp_free() was called when there was nothing allocating, leading to oops. Signed-off-by:
-
Pavel Machek authored
Move device name resolution code around so that it is not called from resume-from-initrd. name_to_dev_t may be unavailable at that point. Signed-off-by:
-
Bernard Blackham authored
Fix u32 vs pm_message_t confusion in cpufreq. Signed-off-by:
Bernard Blackham <bernard@blackham.com.au> Signed-off-by:
-
Pavel Machek authored
Few more u32 vs. pm_message_t fixes. Signed-off-by:
-
Dave Jones authored
If CONFIG_NUMA isn't set, we use the define in <linux/mmzone.h> for early_pfn_to_nid (which defines it to 0). Because of this, the prototype needs to move inside the CONFIG_NUMA too, or anal gcc's get really confused. Signed-off-by:
Dave Jones <davej@redhat.com> Signed-off-by:
-
Dave Jones authored
The recent cleanups to asm-i386/mmzone.h were suboptimal nesting an ifdef of the same symbol. This patch removes some of the ifdef'ery to make things more readable again. Signed-off-by:
-
Shaohua Li authored
There has been some discuss about solving the SMP MTRR suspend/resume breakage, but I didn't find a patch for it. This is an intent for it. The basic idea is moving mtrr initializing into cpu_identify for all APs (so it works for cpu hotplug). For BP, restore_processor_state is responsible for restoring MTRR. Signed-off-by:
Shaohua Li <shaohua.li@intel.com> Acked-by:
Andi Kleen <ak@muc.de> Signed-off-by:
-
Adrian Bunk authored
This patch by Yoshihiro MATSUYAMA (already ACK'ed by David Howells) adds a defconfig for the frv arch. Signed-Off-By:
Yoshihiro MATSUYAMA <y.matsu@jp.fujitsu.com> Signed-off-by:
Adrian Bunk <bunk@stusta.de> Cc: David Howells <dhowells@redhat.com> Signed-off-by:
-
Anton Blanchard authored
We dont need to use the PERFMON exception on POWER5, in fact the firmware returns an error. Due to this just remove the warning. Also now that we have proper runlatch support we can remove the bootup hack. Signed-off-by:
Anton Blanchard <anton@samba.org> Signed-off-by:
-
Michael Ellerman authored
Not sure if we really need this, but it was handy to know which iSeries loop I was testing. Be consistent about printing which idle loop we're using, with this patch we cover all cases. Signed-off-by:
Michael Ellerman <michael@ellerman.id.au> Signed-off-by:
-
Anton Blanchard authored
Fix a compile warning introduced by the previous patches. Signed-off-by:
-
Anton Blanchard authored
- remove some unnecessary includes - add runlatch support - no need to use raw_smp_processor_id any more, current preempt debug logic checks for processes that are bound to one cpu. Signed-off-by:
-
Anton Blanchard authored
- separate out sleep logic in dedicated_idle, it was so far indented that it got squashed against the right side of the screen. - add runlatch support, looping on runlatch disable. Signed-off-by:
-
Anton Blanchard authored
- remove min/max yield time, we dont use the values anywhere - separate shared and dedicated idle loops - check need_resched again with irqs off to avoid sleeping with pending work - continually set runlatch off in idle loop, this means we dont need to turn the runlatch off on exception exit and suffer that associated cost for all exceptions. (A future patch will turn the runlatch on at exception entry) Signed-off-by:
-
Michael Ellerman authored
Now that the idle loop is configured by each platform we don't need idle_setup() anymore. Signed-off-by:
-
Michael Ellerman authored
This patch fixes up iSeries, pSeries, pmac and maple to set the correct idle function for each platform. Signed-off-by:
-
Michael Ellerman authored
dedicated_idle() and shared_idle() are only used by pSeries, so move them into pSeries_setup.c Signed-off-by:
-
Michael Ellerman authored
Move iSeries_idle() into iSeries_setup.c, no one else needs to know about it. Signed-off-by:
-
Michael Ellerman authored
This patch adds an idle member to the ppc_md structure and calls it from cpu_idle(). If a platform leaves ppc_md.idle as null it will get the default idle loop default_idle(). Signed-off-by:
-
Milton Miller authored
Now that hvc_get_chars doesn't strip NULs, hvsi doesn't have to duplicate it. Signed-off-by:
Milton Miller <miltonm@bga.com> Signed-off-by:
-
Milton Miller authored
Separate the NUL character filtering from get_hvc_chars. Signed-off-by:
-
Milton Miller authored
When registering the hvc console port, register a list of ops (read and write) to go with it, instead of calling fixed function names. This allows different ports to encode the data differently. Signed-off-by:
-
Milton Miller authored
Remove all the vio device driver code from hvc_console.c This will allow us to separate hvsi, hvc, and allow hvc_console to be used without the ppc64 vio layer. Signed-off-by:
-
Milton Miller authored
Separate the console setup routines of the hvc_console and the vio layer. Remove the call to find_init_vty from hvc_console.c. Fail the setup routine if the console doesn't exist, but register the console again when the specified channel is instantiated. This scheme maintains the print buffer semantics while eliminating callout and call back for the console code. Signed-off-by:
-
Milton Miller authored
Check if a vterm was registered before accepting it as a console. Check that a slot hasn't been probed with a tty in hvc_instantiate(). Check that a slot hasn't been free'ed when handing out console device. Signed-off-by:
-
Milton Miller authored
Statically initialize the vtermnos array. Signed-off-by:
-
Milton Miller authored
num_vterms hasn't been used since the hotplug support went in. Also, remove a dead code line from a list_for_each_entry conversion. Signed-off-by:
-
Milton Miller authored
hvc_console checks MAGIC_SYSRQ and XMON config vars. Signed-off-by:
-
Milton Miller authored
Be thorough in our exit routine, since it says it is there to be so. Unregistering without registering is safe (checked in 2.6.10). Signed-off-by:
-
Milton Miller authored
Guard the MAGIC_SYSRQ ^O to be just on the console channel. Make the other channels more transparent. Signed-off-by:
-
Milton Miller authored
Have the hvc console code try to pull characters immediately when receiving an interrupt, and kick the poll thread only if the immediate poll indicates it needed a call back to do more work. Signed-off-by:
-
Milton Miller authored
Use the vterm numbers to match the vio devices being probed with the indices already allocated via the console initcall function hvc_find_vtys. The old code required hvc_find_vtys to "guess" the matching devices the vio subsystem would find and its probe order. Signed-off-by:
-
Milton Miller authored
Milton Miller has done a lot of work to clean up our hvc_console code. One of the important things the following patch series does is separate the VIO layer from the hvc_console code. With the VIO specific code removed any ppc64 platform, or even any architecture, can use hvc_console as a generic polling console. You simply have to supply a get_chars and put_chars method and hvc_console does the rest of the work. You can even use it for an interrupt driven console. This patch: Rearrange the code in drivers/char/hvc_console.c to make future patches smaller. No actual code changes, just ordering of the functions in the file. Signed-off-by:
-
Anton Blanchard authored
We already have a prototype for sys_remap_file_pages (239) so there is no need to reserve it twice. Signed-off-by:
-
Anton Blanchard authored
- Clean up sys32_getpriority comment. - Add ioprio syscalls, and sign extend 32bit versions. Signed-off-by:
-
Anton Blanchard authored
Remove some unnecessary includes, an out of date comment and a prototype for sys_timer_create (which is now in syscalls.h) Signed-off-by:
-
Anton Blanchard authored
- Make ioprio syscalls return long, like set/getpriority syscalls. - Move function prototypes into syscalls.h so we can pick them up in the 32/64bit compat code. Signed-off-by:
Jens Axboe <axboe@suse.de> Signed-off-by:
-
Anton Blanchard authored
Enable the runlatch at the start of each exception. Unfortunately we are out of space in the 0x300 handler, so I added it a bit later. The SPR write is fairly expensive, perhaps we should cache the runlatch state in the paca and avoid the write when possible. We don't need to turn the runlatch off, we do that in the idle loop. Better to take the hit in the idle loop than for each exception exit. Signed-off-by:
-
