aufs vs. m68k conflict, please advice
Hi,
a build of linux-2.6 (3.2~rc4-1~experimental.1) with gcc-4.6 (to check whether we can switch to it for the kernel, too) fails: […] LD [M] fs/affs/affs.o LD fs/aufs/built-in.o CC [M] fs/aufs/module.o In file included from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/include/linux/hardirq.h:7:0, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/arch/m68k/include/asm/irqflags.h:6, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/include/linux/irqflags.h:15, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/include/linux/spinlock.h:53, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/include/linux/seqlock.h:29, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/include/linux/time.h:8, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/include/linux/stat.h:60, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/include/linux/module.h:10, from /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/fs/aufs/module.c:23: /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/arch/m68k/include/asm/hardirq.h: In function 'ack_bad_irq': /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/arch/m68k/include/asm/hardirq.h:23:2: error: expected ')' before 'AUFS_NAME' make[7]: *** [fs/aufs/module.o] Error 1 […] The cause isn’t hard to figure out: I /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/arch/m68k R21 <40 C1 384 |180 115|73 14:39 21 static inline void ack_bad_irq(unsigned int irq) 22 { 23 pr_crit("unexpected IRQ trap at vector %02x ", irq); 24 } (pbuild26252)root@ara5:/ # fgrep AUFS_NAME /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/fs/aufs/* /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/fs/aufs/Makefile:ccflags-y += -D'pr_fmt(fmt)=AUFS_NAME" 40%s:%d:%s[%d]: 40"fmt,__func__,__LINE__,current->comm,current->pid' This, to me, looks like cpp abuse in aufs, but I’m not a kernel programmer (Linux or otherwise). bye, //mirabilos -- Support mksh as /bin/sh and RoQA dash NOW! ‣ src:bash (240 (258) bugs: 0 RC, 167 (181) I&N, 73 (77) M&W, 0 F&P) ‣ src:dash (72 (82) bugs: 3 RC, 27 (30) I&N, 42 (49) M&W, 0 F&P) ‣ src:mksh (1 bug: 0 RC, 0 I&N, 1 M&W, 0 F&P) http://qa.debian.org/data/bts/graphs/d/dash.png is pretty red, innit? -- To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org Archive: Pine.BSM.4.64L.1112161437090.856@herc.mirbsd.org"> http://lists.debian.org/Pine.BSM.4.64L.1112161437090.856@herc.mirbsd.org |
aufs vs. m68k conflict, please advice
On Fri, 2011-12-16 at 14:42 +0000, Thorsten Glaser wrote:
> Hi, > > a build of linux-2.6 (3.2~rc4-1~experimental.1) with gcc-4.6 (to > check whether we can switch to it for the kernel, too) fails: [...] > /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/arch/m68k/include/asm/hardirq.h: In function 'ack_bad_irq': > /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/arch/m68k/include/asm/hardirq.h:23:2: error: expected ')' before 'AUFS_NAME' > make[7]: *** [fs/aufs/module.o] Error 1 > […] > > The cause isn’t hard to figure out: > > I /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/arch/m68k R21 <40 C1 384 |180 115|73 14:39 > 21 static inline void ack_bad_irq(unsigned int irq) > 22 { > 23 pr_crit("unexpected IRQ trap at vector %02x ", irq); > 24 } > > (pbuild26252)root@ara5:/ # fgrep AUFS_NAME /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/fs/aufs/* > /tmp/buildd/linux-2.6-3.2~rc4/debian/build/source_m68k_none/fs/aufs/Makefile:ccflags-y += -D'pr_fmt(fmt)=AUFS_NAME" 40%s:%d:%s[%d]: 40"fmt,__func__,__LINE__,current->comm,current->pid' > > This, to me, looks like cpp abuse in aufs, but I’m not a kernel > programmer (Linux or otherwise). Maybe, but it should work just as long as AUFS_NAME is also defined in advance. Not sure why that's not also provided on the command line, or why other architectures get away with it. Maybe they just don't use pr_*() in headers. Ben. -- Ben Hutchings Computers are not intelligent. They only think they are. |
aufs vs. m68k conflict, please advice
Ben Hutchings dixit:
>Maybe, but it should work just as long as AUFS_NAME is also defined in Right, looks like it. >advance. Not sure why that's not also provided on the command line, or Hrm. Can you forward this to the aufs people then, and maybe get us some fix? >why other architectures get away with it. Maybe they just don't use >pr_*() in headers. *shrug* Like I said, no idea. I’m “just” the build “dæmon” ☺ (ok, with a bit of knowledge but nothing really learned) bye, //mirabilos -- <ch> you introduced a merge commit │<mika> % g rebase -i HEAD^^ <mika> sorry, no idea and rebasing just fscked │<mika> Segmentation <ch> should have cloned into a clean repo │ fault (core dumped) <ch> if I rebase that now, it's really ugh │<mika:#grml> wuahhhhhh -- To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org Archive: Pine.BSM.4.64L.1112161714000.856@herc.mirbsd.org"> http://lists.debian.org/Pine.BSM.4.64L.1112161714000.856@herc.mirbsd.org |
aufs vs. m68k conflict, please advice
Ben Hutchings dixit:
>why other architectures get away with it. Maybe they just don't use >pr_*() in headers. Maybe something like this? #define ack_bad_irq(irq) do { pr_crit("unexpected IRQ trap at vector %02x ", (unsigned int)(irq)); } while (/* CONSTCOND */ 0) This would defer pr_crit expansion to when the static inline function was actually used. Just an idea of the moment, //mirabilos -- “Having a smoking section in a restaurant is like having a peeing section in a swimming pool.” -- Edward Burr -- To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org Archive: Pine.BSM.4.64L.1112171426140.856@herc.mirbsd.org"> http://lists.debian.org/Pine.BSM.4.64L.1112171426140.856@herc.mirbsd.org |
aufs vs. m68k conflict, please advice
Dixi quod…
>Maybe something like this? With that, aufs indeed compiles and module-links. bye, //mirabilos -- “Having a smoking section in a restaurant is like having a peeing section in a swimming pool.” -- Edward Burr -- To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org Archive: Pine.BSM.4.64L.1112171623570.856@herc.mirbsd.org"> http://lists.debian.org/Pine.BSM.4.64L.1112171623570.856@herc.mirbsd.org |
aufs vs. m68k conflict, please advice
On Sat, 2011-12-17 at 14:28 +0000, Thorsten Glaser wrote:
> Ben Hutchings dixit: > > >why other architectures get away with it. Maybe they just don't use > >pr_*() in headers. > > Maybe something like this? > > #define ack_bad_irq(irq) do { > pr_crit("unexpected IRQ trap at vector %02x ", > (unsigned int)(irq)); > } while (/* CONSTCOND */ 0) > > This would defer pr_crit expansion to when the static inline > function was actually used. No, I don't think this is a bug in the header. Ben. -- Ben Hutchings Computers are not intelligent. They only think they are. |
aufs vs. m68k conflict, please advice
On Sat, Dec 17, 2011 at 02:28:35PM +0000, Thorsten Glaser wrote:
> Ben Hutchings dixit: > > >why other architectures get away with it. Maybe they just don't use > >pr_*() in headers. > > Maybe something like this? > > #define ack_bad_irq(irq) do { > pr_crit("unexpected IRQ trap at vector %02x ", > (unsigned int)(irq)); > } while (/* CONSTCOND */ 0) > > This would defer pr_crit expansion to when the static inline > function was actually used. > > Just an idea of the moment, IMHO the problem is that aufs provides an incomplete definition of pr_fmt. Either it should define AUFS_NAME on the commandline, too, or should define pr_fmt in an aufs header (or a .c file) #included after all other headers and only when AUFS_NAME is defined, too. The ugly thing about aufs' pr_fmt being already there when ack_bad_irq is defined is, that the message printed by the pr_crit suddenly looks aufs specific which it clearly isn't. So it should better make sure that the definition isn't available to ack_bad_irq. Best regards Uwe -- Pengutronix e.K. | Uwe Kleine-Knig | Industrial Linux Solutions | http://www.pengutronix.de/ | -- To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org Archive: 20111217184915.GL24496@pengutronix.de">http://lists.debian.org/20111217184915.GL24496@pengutronix.de |
aufs vs. m68k conflict, please advice
Uwe Kleine-K�nig dixit:
>On Sat, Dec 17, 2011 at 02:28:35PM +0000, Thorsten Glaser wrote: >> Maybe something like this? […] >> Just an idea of the moment, Well, it does make the thing compile with minimal effort. >IMHO the problem is that aufs provides an incomplete definition of >pr_fmt. Either it should define AUFS_NAME on the commandline, too, or >should define pr_fmt in an aufs header (or a .c file) #included after >all other headers and only when AUFS_NAME is defined, too. My initial thoughts, too. >The ugly thing about aufs' pr_fmt being already there when ack_bad_irq >is defined is, that the message printed by the pr_crit suddenly looks >aufs specific which it clearly isn't. So it should better make sure that >the definition isn't available to ack_bad_irq. True, but looking at the actual changes, it doesn’t look too aufs specific to me. (If the function ack_bad_irq is instantiated in the aufs code at all, which is debatable; a quick fgrep -r doesn’t find anything.) Anyway, will please someone communicate that to the aufs developers and include some sort of fix in the next upload? (My build is still compiling, 132 bogomips is fast; will communicate success or failure of the kernel in general.) Thanks, //mirabilos -- <dileks> ch: good, you corrected yourself. ppl tend to tweet such news immediately, sth. like "grml devs seem to be buyable" <ch> dileks: we _are_. if you throw enough money in our direction, things will happen <mika> everyone is buyable, it's just a matter of price <mrud> and now comes [mira] and uses this as a signature ;0 -- they asked for it… -- To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org Archive: Pine.BSM.4.64L.1112171856380.856@herc.mirbsd.org"> http://lists.debian.org/Pine.BSM.4.64L.1112171856380.856@herc.mirbsd.org |
aufs vs. m68k conflict, please advice
On Sat, Dec 17, 2011 at 07:00:00PM +0000, Thorsten Glaser wrote:
> Uwe Kleine-K�nig dixit: Your mailer is broken. And it added: X-Message-Flag: Your mailer is broken. Get an update at http://www.washington.edu/pine/getpine/pcpine.html for free. to your mail :-) I guess it means you not me? > >On Sat, Dec 17, 2011 at 02:28:35PM +0000, Thorsten Glaser wrote: > > >> Maybe something like this? > […] > >> Just an idea of the moment, > > Well, it does make the thing compile with minimal effort. > > >IMHO the problem is that aufs provides an incomplete definition of > >pr_fmt. Either it should define AUFS_NAME on the commandline, too, or > >should define pr_fmt in an aufs header (or a .c file) #included after > >all other headers and only when AUFS_NAME is defined, too. > > My initial thoughts, too. > > >The ugly thing about aufs' pr_fmt being already there when ack_bad_irq > >is defined is, that the message printed by the pr_crit suddenly looks > >aufs specific which it clearly isn't. So it should better make sure that > >the definition isn't available to ack_bad_irq. > > True, but looking at the actual changes, it doesn’t look too aufs > specific to me. (If the function ack_bad_irq is instantiated in > the aufs code at all, which is debatable; a quick fgrep -r doesn’t > find anything.) Right, probably pr_something just shouldn't be used in headers I think. The easiest fix would be: -ccflags-y += -D'pr_fmt(fmt)=AUFS_NAME" 40%s:%d:%s[%d]: 40"fmt,__func__,__LINE__,current->comm,current->pid' +ccflags-y += -D'pr_fmt(fmt)="aufs 40%s:%d:%s[%d]: 40"fmt,__func__,__LINE__,current->comm,current->pid' But as it is ugly to have that in a Makefile, you can also try the patch below. Best regards Uwe >From ad5d7fd2630feea17a1a6fffac74fdcb0505b2ee Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Uwe=20Kleine-K=C3=B6nig?= <u.kleine-koenig@pengutronix.de> Date: Sat, 17 Dec 2011 20:45:21 +0100 Subject: [PATCH] new patch to not leak pr_fmt to foreign headers --- patches/features/all/aufs3/dont-leak-pr_fmt.patch | 211 +++++++++++++++++++++ patches/series/base | 1 + 2 files changed, 212 insertions(+), 0 deletions(-) create mode 100644 patches/features/all/aufs3/dont-leak-pr_fmt.patch diff --git a/patches/features/all/aufs3/dont-leak-pr_fmt.patch b/patches/features/all/aufs3/dont-leak-pr_fmt.patch new file mode 100644 index 0000000..93e05ca --- /dev/null +++ b/patches/features/all/aufs3/dont-leak-pr_fmt.patch @@ -0,0 +1,211 @@ +From: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> +Subject: [PATCH] don't leak incomplete definition of pr_fmt to headers + +This fixes a build problem on m68k, more details can be found at: +http://lists.debian.org/debian-kernel/2011/12/msg00460.html +--- + fs/aufs/Makefile | 2 -- + fs/aufs/aufs.h | 4 ++++ + fs/aufs/branch.h | 1 - + fs/aufs/cpup.h | 1 - + fs/aufs/dbgaufs.h | 1 - + fs/aufs/debug.h | 1 - + fs/aufs/dentry.h | 1 - + fs/aufs/dir.h | 1 - + fs/aufs/dynop.h | 1 - + fs/aufs/file.h | 1 - + fs/aufs/fstype.h | 1 - + fs/aufs/inode.h | 1 - + fs/aufs/opts.h | 1 - + fs/aufs/rdu.c | 1 - + fs/aufs/super.h | 1 - + fs/aufs/sysaufs.h | 1 - + fs/aufs/whout.h | 1 - + fs/aufs/wkq.h | 1 - + 18 files changed, 4 insertions(+), 18 deletions(-) + +--- source_amd64_none.orig/fs/aufs/Makefile ++++ source_amd64_none/fs/aufs/Makefile +@@ -8,8 +8,6 @@ + # cf. include/linux/kernel.h + # enable pr_debug + ccflags-y += -DDEBUG +-# sparse doesn't allow spaces +-ccflags-y += -D'pr_fmt(fmt)=AUFS_NAME" 40%s:%d:%s[%d]: 40"fmt,__func__,__LINE__,current->comm,current->pid' + + obj-$(CONFIG_AUFS_FS) += aufs.o + aufs-y := module.o sbinfo.o super.o branch.o xino.o sysaufs.o opts.o +--- source_amd64_none.orig/fs/aufs/aufs.h ++++ source_amd64_none/fs/aufs/aufs.h +@@ -33,6 +33,11 @@ + #define AuStubInt0(name, ...) + AuStub(int, name, return 0, __VA_ARGS__) + ++#include <linux/aufs_type.h> ++ ++#undef pr_fmt ++#define pr_fmt(fmt) AUFS_NAME " %s:%d:%s[%d]: " fmt, __func__, __LINE__, current->comm, current->pid ++ + #include "debug.h" + + #include "branch.h" +--- source_amd64_none.orig/fs/aufs/debug.h ++++ source_amd64_none/fs/aufs/debug.h +@@ -35,7 +35,6 @@ + #include <linux/delay.h> + /* #include <linux/kd.h> */ + #include <linux/sysrq.h> +-#include <linux/aufs_type.h> + + #include <asm/system.h> + +--- source_amd64_none.orig/fs/aufs/file.h ++++ source_amd64_none/fs/aufs/file.h +@@ -28,7 +28,6 @@ + #include <linux/file.h> + #include <linux/fs.h> + #include <linux/poll.h> +-#include <linux/aufs_type.h> + #include "rwsem.h" + + struct au_branch; +--- source_amd64_none.orig/fs/aufs/inode.h ++++ source_amd64_none/fs/aufs/inode.h +@@ -27,7 +27,6 @@ + + #include <linux/fs.h> + #include <linux/fsnotify.h> +-#include <linux/aufs_type.h> + #include "rwsem.h" + + struct vfsmount; +--- source_amd64_none.orig/fs/aufs/rdu.c ++++ source_amd64_none/fs/aufs/rdu.c +@@ -24,7 +24,6 @@ + #include <linux/fs_stack.h> + #include <linux/security.h> + #include <linux/uaccess.h> +-#include <linux/aufs_type.h> + #include "aufs.h" + + /* bits for struct aufs_rdu.flags */ +--- source_amd64_none.orig/fs/aufs/branch.h ++++ source_amd64_none/fs/aufs/branch.h +@@ -27,7 +27,6 @@ + + #include <linux/fs.h> + #include <linux/mount.h> +-#include <linux/aufs_type.h> + #include "dynop.h" + #include "rwsem.h" + #include "super.h" +--- source_amd64_none.orig/fs/aufs/cpup.h ++++ source_amd64_none/fs/aufs/cpup.h +@@ -27,7 +27,6 @@ + + #include <linux/path.h> + #include <linux/time.h> +-#include <linux/aufs_type.h> + + struct inode; + struct file; +--- source_amd64_none.orig/fs/aufs/dbgaufs.h ++++ source_amd64_none/fs/aufs/dbgaufs.h +@@ -26,7 +26,6 @@ + #ifdef __KERNEL__ + + #include <linux/init.h> +-#include <linux/aufs_type.h> + + struct super_block; + struct au_sbinfo; +--- source_amd64_none.orig/fs/aufs/dentry.h ++++ source_amd64_none/fs/aufs/dentry.h +@@ -26,7 +26,6 @@ + #ifdef __KERNEL__ + + #include <linux/dcache.h> +-#include <linux/aufs_type.h> + #include "rwsem.h" + + struct au_hdentry { +--- source_amd64_none.orig/fs/aufs/dir.h ++++ source_amd64_none/fs/aufs/dir.h +@@ -26,7 +26,6 @@ + #ifdef __KERNEL__ + + #include <linux/fs.h> +-#include <linux/aufs_type.h> + + /* ---------------------------------------------------------------------- */ + +--- source_amd64_none.orig/fs/aufs/dynop.h ++++ source_amd64_none/fs/aufs/dynop.h +@@ -28,7 +28,6 @@ + #include <linux/fs.h> + #include <linux/mm.h> + #include <linux/rcupdate.h> +-#include <linux/aufs_type.h> + #include "inode.h" + + enum {AuDy_AOP, AuDyLast}; +--- source_amd64_none.orig/fs/aufs/fstype.h ++++ source_amd64_none/fs/aufs/fstype.h +@@ -28,7 +28,6 @@ + #include <linux/fs.h> + #include <linux/magic.h> + #include <linux/romfs_fs.h> +-#include <linux/aufs_type.h> + + static inline int au_test_aufs(struct super_block *sb) + { +--- source_amd64_none.orig/fs/aufs/opts.h ++++ source_amd64_none/fs/aufs/opts.h +@@ -26,7 +26,6 @@ + #ifdef __KERNEL__ + + #include <linux/path.h> +-#include <linux/aufs_type.h> + + struct file; + struct super_block; +--- source_amd64_none.orig/fs/aufs/super.h ++++ source_amd64_none/fs/aufs/super.h +@@ -26,7 +26,6 @@ + #ifdef __KERNEL__ + + #include <linux/fs.h> +-#include <linux/aufs_type.h> + #include "rwsem.h" + #include "spl.h" + #include "wkq.h" +--- source_amd64_none.orig/fs/aufs/sysaufs.h ++++ source_amd64_none/fs/aufs/sysaufs.h +@@ -26,7 +26,6 @@ + #ifdef __KERNEL__ + + #include <linux/sysfs.h> +-#include <linux/aufs_type.h> + #include "module.h" + + struct super_block; +--- source_amd64_none.orig/fs/aufs/whout.h ++++ source_amd64_none/fs/aufs/whout.h +@@ -25,7 +25,6 @@ + + #ifdef __KERNEL__ + +-#include <linux/aufs_type.h> + #include "dir.h" + + /* whout.c */ +--- source_amd64_none.orig/fs/aufs/wkq.h ++++ source_amd64_none/fs/aufs/wkq.h +@@ -28,7 +28,6 @@ + + #include <linux/sched.h> + #include <linux/wait.h> +-#include <linux/aufs_type.h> + + struct super_block; + diff --git a/patches/series/base b/patches/series/base index dc8b037..98916b1 100644 --- a/patches/series/base +++ b/patches/series/base @@ -14,6 +14,7 @@ + features/all/aufs3/aufs3-add.patch # mark as staging/crap + features/all/aufs3/mark-as-staging.patch ++ features/all/aufs3/dont-leak-pr_fmt.patch + bugfix/ia64/hardcode-arch-script-output.patch + bugfix/mips/disable-advansys.patch -- 1.7.7.3 -- Pengutronix e.K. | Uwe Kleine-König | Industrial Linux Solutions | http://www.pengutronix.de/ | -- To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org Archive: 20111217195719.GM24496@pengutronix.de">http://lists.debian.org/20111217195719.GM24496@pengutronix.de |
| All times are GMT. The time now is 07:34 PM. |
VBulletin, Copyright ©2000 - 2013, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.