linaro-dev March 2011

linaro-dev@lists.linaro.org

123 participants
172 discussions

[PATCH] arm: handle ptrace requests to change PC during interrupted system calls

by Arnd Bergmann

From: Ulrich Weigand <ulrich.weigand(a)linaro.org> GDB's interrupt.exp test cases currenly fail on ARM. The problem is how do_signal handled restarting interrupted system calls: The entry.S assembler code determines that we come from a system call; and that information is passed as "syscall" parameter to do_signal. That routine then calls get_signal_to_deliver [*] and if a signal is to be delivered, calls into handle_signal. If a system call is to be restarted either after the signal handler returns, or if no handler is to be called in the first place, the PC is updated after the get_signal_to_deliver call, either in handle_signal (if we have a handler) or at the end of do_signal (otherwise). Now the problem is that during [*], the call to get_signal_to_deliver, a ptrace intercept may happen. During this intercept, the debugger may change registers, including the PC. This is done by GDB if it wants to execute an "inferior call", i.e. the execution of some code in the debugged program triggered by GDB. To this purpose, GDB will save all registers, allocate a stack frame, set up PC and arguments as appropriate for the call, and point the link register to a dummy breakpoint instruction. Once the process is restarted, it will execute the call and then trap back to the debugger, at which point GDB will restore all registers and continue original execution. This generally works fine. However, now consider what happens when GDB attempts to do exactly that while the process was interrupted during execution of a to-be- restarted system call: do_signal is called with the syscall flag set; it calls get_signal_to_deliver, at which point the debugger takes over and changes the PC to point to a completely different place. Now get_signal_to_deliver returns without a signal to deliver; but now do_signal decides it should be restarting a system call, and decrements the PC by 2 or 4 -- so it now points to 2 or 4 bytes before the function GDB wants to call -- which leads to a subsequent crash. To fix this problem, two things need to be supported: - do_signal must be able to recognize that get_signal_to_deliver changed the PC to a different location, and skip the restart-syscall sequence - once the debugger has restored all registers at the end of the inferior call sequence, do_signal must recognize that *now* it needs to restart the pending system call, even though it was now entered from a breakpoint instead of an actual svc instruction This set of issues is solved on other platforms, usually by one of two mechanisms: - The status information "do_signal is handling a system call that may need restarting" is itself carried in some register that can be accessed via ptrace. This is e.g. on Intel the "orig_eax" register; on Sparc the kernel defines a magic extra bit in the flags register for this purpose. This allows GDB to manage that state: reset it when doing an inferior call, and restore it after the call is finished. - On s390, do_signal transparently handles this problem without requiring GDB interaction, by performing system call restarting in the following way: first, adjust the PC as necessary for restarting the call. Then, call get_signal_to_deliver; and finally just continue execution at the PC. This way, if GDB does not change the PC, everything is as before. If GDB *does* change the PC, execution will simply continue there -- and once GDB restores the PC it saved at that point, it will automatically point to the *restarted* system call. (There is the minor twist how to handle system calls that do *not* need restarting -- do_signal will undo the PC change in this case, after get_signal_to_deliver has returned, and only if ptrace did not change the PC during that call.) Because there does not appear to be any obvious register to carry the syscall-restart information on ARM, we'd either have to introduce a new artificial ptrace register just for that purpose, or else handle the issue transparently like on s390. The patch below implements the second option; using this patch makes the interrupt.exp test cases pass on ARM, with no regression in the GDB test suite otherwise. Signed-off-by: Ulrich Weigand <ulrich.weigand(a)linaro.org> Signed-off-by: Arnd Bergmann <arnd.bergmann(a)linaro.org> Cc: patches(a)linaro.org --- arch/arm/kernel/signal.c | 90 +++++++++++++++++++++++++++++++++++++-------------------------- 1 file changed, 53 insertions(+), 37 deletions(-) diff --git a/arch/arm/kernel/signal.c b/arch/arm/kernel/signal.c index cb83983..8242197 100644 --- a/arch/arm/kernel/signal.c +++ b/arch/arm/kernel/signal.c @@ -597,19 +597,13 @@ setup_rt_frame(int usig, struct k_sigaction *ka, siginfo_t *info, return err; } -static inline void setup_syscall_restart(struct pt_regs *regs) -{ - regs->ARM_r0 = regs->ARM_ORIG_r0; - regs->ARM_pc -= thumb_mode(regs) ? 2 : 4; -} - /* * OK, we're invoking a handler */ static int handle_signal(unsigned long sig, struct k_sigaction *ka, siginfo_t *info, sigset_t *oldset, - struct pt_regs * regs, int syscall) + struct pt_regs * regs) { struct thread_info *thread = current_thread_info(); struct task_struct *tsk = current; @@ -617,26 +611,6 @@ handle_signal(unsigned long sig, struct k_sigaction *ka, int ret; /* - * If we were from a system call, check for system call restarting... - */ - if (syscall) { - switch (regs->ARM_r0) { - case -ERESTART_RESTARTBLOCK: - case -ERESTARTNOHAND: - regs->ARM_r0 = -EINTR; - break; - case -ERESTARTSYS: - if (!(ka->sa.sa_flags & SA_RESTART)) { - regs->ARM_r0 = -EINTR; - break; - } - /* fallthrough */ - case -ERESTARTNOINTR: - setup_syscall_restart(regs); - } - } - - /* * translate the signal */ if (usig < 32 && thread->exec_domain && thread->exec_domain->signal_invmap) @@ -685,6 +659,7 @@ handle_signal(unsigned long sig, struct k_sigaction *ka, */ static void do_signal(struct pt_regs *regs, int syscall) { + unsigned int retval = 0, continue_addr = 0, restart_addr = 0; struct k_sigaction ka; siginfo_t info; int signr; @@ -698,18 +673,61 @@ static void do_signal(struct pt_regs *regs, int syscall) if (!user_mode(regs)) return; + /* + * If we were from a system call, check for system call restarting... + */ + if (syscall) { + continue_addr = regs->ARM_pc; + restart_addr = continue_addr - (thumb_mode(regs) ? 2 : 4); + retval = regs->ARM_r0; + + /* + * Prepare for system call restart. We do this here so that a + * debugger will see the already changed PSW. + */ + switch (retval) { + case -ERESTARTNOHAND: + case -ERESTARTSYS: + case -ERESTARTNOINTR: + regs->ARM_r0 = regs->ARM_ORIG_r0; + regs->ARM_pc = restart_addr; + break; + case -ERESTART_RESTARTBLOCK: + regs->ARM_r0 = -EINTR; + break; + } + } + if (try_to_freeze()) goto no_signal; + /* + * Get the signal to deliver. When running under ptrace, at this + * point the debugger may change all our registers ... + */ signr = get_signal_to_deliver(&info, &ka, regs, NULL); if (signr > 0) { sigset_t *oldset; + /* + * Depending on the signal settings we may need to revert the + * decision to restart the system call. But skip this if a + * debugger has chosen to restart at a different PC. + */ + if (regs->ARM_pc == restart_addr) { + if (retval == -ERESTARTNOHAND + || (retval == -ERESTARTSYS + && !(ka.sa.sa_flags & SA_RESTART))) { + regs->ARM_r0 = -EINTR; + regs->ARM_pc = continue_addr; + } + } + if (test_thread_flag(TIF_RESTORE_SIGMASK)) oldset = &current->saved_sigmask; else oldset = &current->blocked; - if (handle_signal(signr, &ka, &info, oldset, regs, syscall) == 0) { + if (handle_signal(signr, &ka, &info, oldset, regs) == 0) { /* * A signal was successfully delivered; the saved * sigmask will have been stored in the signal frame, @@ -723,11 +741,14 @@ static void do_signal(struct pt_regs *regs, int syscall) } no_signal: - /* - * No signal to deliver to the process - restart the syscall. - */ if (syscall) { - if (regs->ARM_r0 == -ERESTART_RESTARTBLOCK) { + /* + * Handle restarting a different system call. As above, + * if a debugger has chosen to restart at a different PC, + * ignore the restart. + */ + if (retval == -ERESTART_RESTARTBLOCK + && regs->ARM_pc == continue_addr) { if (thumb_mode(regs)) { regs->ARM_r7 = __NR_restart_syscall - __NR_SYSCALL_BASE; regs->ARM_pc -= 2; @@ -750,11 +771,6 @@ static void do_signal(struct pt_regs *regs, int syscall) #endif } } - if (regs->ARM_r0 == -ERESTARTNOHAND || - regs->ARM_r0 == -ERESTARTSYS || - regs->ARM_r0 == -ERESTARTNOINTR) { - setup_syscall_restart(regs); - } /* If there's no signal to deliver, we just put the saved sigmask * back.

14 years, 10 months

Engineering Resources news

by Matt Waddel

Last week a new tab was added to the banner of wiki.linaro.org called "HowTo". This links to a page we have been working on that is intended to point people to some of our most common how-to scenarios. In addition it includes a pointer to pages that are tagged with "CategoryHowTo", just click on the "list of all how-to's" link. We hope that promoting the visibility of this page will make it easier for others to find the Linaro information they are looking for. Please take a look at the page when you get some time: https://wiki.linaro.org/Resources/HowTo There might be a link you'll find helpful. Also, please let us know if you think there is an existing page in the wiki that we should add to this page.

14 years, 10 months

[PATCH android/system/vold] Promote the max partition support to 8

by Jeremy Chang

This is needed for supporting multiple partition layout. Change-Id: Ic4e1b5f7219cb51f02d9e0d2c14046da9f7b51f1 Signed-off-by: Jeremy Chang <jeremy.chang(a)linaro.org> --- DirectVolume.h | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/DirectVolume.h b/DirectVolume.h index 4bf14ff..345de09 100644 --- a/DirectVolume.h +++ b/DirectVolume.h @@ -27,7 +27,7 @@ typedef android::List<char *> PathCollection; class DirectVolume : public Volume { public: - static const int MAX_PARTITIONS = 4; + static const int MAX_PARTITIONS = 8; protected: PathCollection *mPaths; int mDiskMajor; -- 1.7.1

14 years, 10 months

[PATCH android/device/linaro/common] Let vold mount sdcard partition

by Jeremy Chang

Create the directories that vold needs to complete the mounting. vold communicates with MountService which manages Android platform storage. Let vold do the sdcard partition mounting so sdcard partition can be the Android external storage. Signed-off-by: Jeremy Chang <jeremy.chang(a)linaro.org> --- init.rc | 15 +++++++-------- 1 files changed, 7 insertions(+), 8 deletions(-) diff --git a/init.rc b/init.rc index af47666..7f94a21 100644 --- a/init.rc +++ b/init.rc @@ -45,22 +45,22 @@ loglevel 3 mkdir /config 0500 root root # Directory for putting things only root should see. - #mkdir /mnt/secure 0700 root root + mkdir /mnt/secure 0700 root root # Directory for staging bindmounts - #mkdir /mnt/secure/staging 0700 root root + mkdir /mnt/secure/staging 0700 root root # Directory-target for where the secure container # imagefile directory will be bind-mounted - #mkdir /mnt/secure/asec 0700 root root + mkdir /mnt/secure/asec 0700 root root # Secure container public mount points. - #mkdir /mnt/asec 0700 root system - #mount tmpfs tmpfs /mnt/asec mode=0755,gid=1000 + mkdir /mnt/asec 0700 root system + mount tmpfs tmpfs /mnt/asec mode=0755,gid=1000 # Filesystem image public mount points. - #mkdir /mnt/obb 0700 root system - #mount tmpfs tmpfs /mnt/obb mode=0755,gid=1000 + mkdir /mnt/obb 0700 root system + mount tmpfs tmpfs /mnt/obb mode=0755,gid=1000 write /proc/sys/kernel/panic_on_oops 1 write /proc/sys/kernel/hung_task_timeout_secs 0 @@ -95,7 +95,6 @@ on fs mount ext4 mmc@blk0p3 /system ro remount mount ext4 mmc@blk0p5 /cache mount ext4 mmc@blk0p6 /data - mount ext4 mmc@blk0p7 /sdcard on post-fs # once everything is setup, no need to modify / -- 1.7.1

14 years, 10 months

[REPORT] 2011-03-17 Engineering Resources team weekly report

by Matt Waddel

Hello Everybody, The Engineering Resources team weekly report for 2011-3-17 can be found at: https://wiki.linaro.org/Resources/Status/WeeklyMeeting/2011-03-17 Highlights below. Best regards, Matt === Action Items from This Meeting === * Improve the Linaro planning structure guides. ie: * engineering BP versus TR BP, when to write a spec * what's a good quality BP * Linaro at UDS introduction for new employees * what are sessions supposed to be about? you need to know what you are going to do *before* you show up. but not too far. * what to expect at the summit * Firewall issues. Use VM's to hack around some of the issues * IRC setup help === Action Items from Last Meeting === * Action items from last meeting with tech leads * AddNewHardware and HardwarePack pages need to explain kernel packages as Steve noted * HardwarePacks page refers to Maverick in stead of Natty - Done * Mentoring section needs to be renamed and back links should be fixed - Done * A better header for wiki that includes things like a "Resources" link in banner/header - Done * Toolchain documentation improvements

14 years, 10 months

Booting Integrator/AP with ARM920T

by Linus Walleij

I have this Integrator/AP development board here in my office: Initialising Boot Monitor System Information Block boot Monitor > i ARM bootPROM [Version 1.3] Rebuilt on Jun 26 2001 at 22:04:10 Running on a Integrator Evaluation Board Board Revision V1.0, ARM920T Processor Memory Size is 128MBytes, Flash Size is 32MBytes Copyright (c) ARM Limited 1999 - 2001. All rights reserved. Board designed by ARM Limited Hardware support provided at http://www.arm.com/ For help on the available commands type ? or h boot Monitor > Does anyone beside me *ever* try to actually boot a recent kernel on this thing? How do you do it? Do I need patched U-boots etc, or can I just load a Z-record image of vmlinux into RAM over the serial console? Needless to say, ARM Ltd. has deleted the web pages for official support of this thing as far as I can see. I was going to try out a few patches affecting mach-integrator... Yours, Linus Walleij

14 years, 10 months

Notes and actions from MM WG call on 15th March

by Sachin Gupta

> > Hi All, > > Please find enclosed link to minutes and actions for multimedia wg > meeting on 15st March 2011. > > > https://wiki.linaro.org/WorkingGroups/Middleware/Multimedia/Notes/2011-03-15 > > Summary > - Kan created seperate AAC encoder lib from Android aac encoder.Talking to > sourceforge opencore-amr maintainers to see if they are willing to maintain > opencore-aac encoder as well > - Bufferpool implementation in xvimagesink patch sent for review to Wim.No > feedback yet > - System metrics app and integrated instrumented player packaging on going > > > Thanks > Sachin >

14 years, 10 months

[PATCH 0/5] make sdhci device drivers self registered

by Shawn Guo

This patch set is to take sdhci device driver specific things out from sdhci-pltfm.c and make them self registered. Here are the difference it makes. * Get the sdhci device driver follow the Linux trend that driver take the registration by its own * sdhci-pltfm.c becomes significantly simpler and only has common support functions there * All sdhci device specific stuff are going back its own driver * The dt and non-dt support share the same pair of .probe and .remove hooks. The first one patch adds common support function into sdhci-pltfm.c when changing sdhci-esdhc-imx driver, while the last one clean up sdhci-pltfm.c when changing sdhci-tegra driver. Only the patch of sdhci-esdhc-imx are tested on hardware, and others are just build tested. Regards, Shawn Shawn Guo (5): mmc: sdhci: make sdhci-esdhc-imx driver self registered mmc: sdhci: make sdhci-cns3xxx driver self registered mmc: sdhci: make sdhci-dove driver self registered mmc: sdhci: make sdhci-tegra driver self registered mmc: sdhci: update Makefile/Kconfig for sdhci_pltfm change drivers/mmc/host/Kconfig | 24 +++-- drivers/mmc/host/Makefile | 11 +- drivers/mmc/host/sdhci-cns3xxx.c | 68 +++++++++++++- drivers/mmc/host/sdhci-dove.c | 70 +++++++++++++- drivers/mmc/host/sdhci-esdhc-imx.c | 98 +++++++++++++++---- drivers/mmc/host/sdhci-pltfm.c | 165 ++----------------------------- drivers/mmc/host/sdhci-pltfm.h | 14 ++- drivers/mmc/host/sdhci-tegra.c | 189 +++++++++++++++++++++---------------

14 years, 10 months

No support of platform device instance id?

by Shawn Guo

Hello all, I'm looking at function of_device_alloc in drivers/of/platform.c, and surprisingly found that the platform device instance id is being hard-coded as -1 when calling platform_device_alloc. Does this mean that there is no support of pdev id in DT? If yes, can anyone please help me understand why we do not have something in DT node to reflect this id and get it supported? Or this is something on TODO list? I'm asking this because I feel we still need pdev id support in DT when reviewing the patch set of 'mx51 basic DT support'. -- Regards, Shawn

14 years, 10 months

Some GL benchmarks for ARM devices

by Jesse Barker

Hi all, Thought this might be of interest to folks. http://www.glbenchmark.com/result.jsp?benchmark=glpro20&orderby=405&screen-… http://www.glbenchmark.com/compare.jsp?benchmark=glpro20&showhide=true&cert… The first is an overall comparison (pretty high level) of available boards, and the second is a breakdown of the top 2 results: the Apple iPad2 and the Hardkernel ODROID-A (FWIW, according to androidtablets.net, the odroid-a is an Orion). While this can certainly be a red herring, it might be nice to see these comparisons with identical screen resolutions (i.e. not all performance measurements are sensitive to pixel resolution). cheers, Jesse

14 years, 10 months

← Newer
1
...
5
6
7
8
9
10
11
...
18
Older →

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

linaro-dev March 2011