Bug#766635: kernel patch for hppa architecture
Source: kernel
Version: 3.16.5-1
Severity: important
Tags: patch
Dear debian kernel maintainers,
Can you please apply this hppa-arch-specific patch to the debian kernel 3.16.5 and keep it until you upgrade to sources of upstream kernel 3.18 ?
Main reason for this patch is to make it possible to use systemd on hppa.
Without this patch people who will by mistake install systemd (e.g. because of dependencies) will render their machines unbootable.
The patch breaks the ABI on hppa, but in a way which will not affect people, because it changes the signals which are usually not used. This has been tested by installing and booting mixtures of glibc and kernel with corresponding patches.
Since this patch affects kernel and glibc, I've opened this debian bug report for the glibc patch:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=766605
Upstream Linux kernel commit (committed into kernel 3.18, as attached here) is 1f25df2eff5b25f52c139d3ff31bc883eee9a0ab
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=1f25df2eff5b25f52c139d3ff31bc883eee9a0ab&utm_source=anzwix
Upstream glibc commit is:
https://sourceware.org/git/?p=glibc.git;a=commit;h=13d845549e41823e6658122dcf268154bcbbcfde
To better explain what we fix here, the glibc changelog description of Carlos is probably best (copied in here):
This is a conscious ABI break for hppa.
We find ourselves unable to run systemd because it expects
SIGRTMIN+29 signals to be available and with hppa starting
at 37 that exceeds the 64 signals available. It is arguable
that the systemd code could compact their signal usage (the
have a gap and don't check SIGRTMAX to see if they are over).
However, that would require pursuing this upstream with systemd.
The least work option is to make hppa more like other arches.
The best option is to free up 3 signals for use with SIGXCPU,
SIGXFSZ and SIGSTKFLT, and move those below 31. We make SIGSYS
equal to SIGUNUSED as is expected. We remove SIGEMT and SIGLOST
as HPUX signal we won't ever use. With that change we match all
other machines.
Given that these signals are so esoteric, testing by other users
building minimal systems from scratch showed no problems. In fact
Tcl fails to build if you make SIGEMT == SIGABRT, so we just removed
SIGEMT (they use a large switch statement in C to handle signals, and
I don't think it's valid to assume they will all have distinct values
to fit into a switch).
Committed as the only solution we possibly have here.
commit 1f25df2eff5b25f52c139d3ff31bc883eee9a0ab
Author: Helge Deller <deller@gmx.de>
Date: Fri Oct 10 22:20:17 2014 +0200
parisc: Reduce SIGRTMIN from 37 to 32 to behave like other Linux architectures
This patch reduces the value of SIGRTMIN on PARISC from 37 to 32, thus
increasing the number of available RT signals and bring it in sync with other
Linux architectures.
Historically we wanted to natively support HP-UX 32bit binaries with the
PA-RISC Linux port. Because of that we carried the various available signals
from HP-UX (e.g. SIGEMT and SIGLOST) and folded them in between the native
Linux signals. Although this was the right decision at that time, this
required us to increase SIGRTMIN to at least 37 which left us with 27 (64-37)
RT signals.
Those 27 RT signals haven't been a problem in the past, but with the upcoming
importance of systemd we now got the problem that systemd alloctes (hardcoded)
signals up to SIGRTMIN+29 which is beyond our NSIG of 64. Because of that we
have not been able to use systemd on the PARISC Linux port yet.
Of course we could ask the systemd developers to not use those hardcoded
values, but this change is very unlikely, esp. with PA-RISC being a niche
architecture.
The other possibility would be to increase NSIG to e.g. 128, but this would
mean to duplicate most of the existing Linux signal handling code into the
parisc specific Linux kernel tree which would most likely introduce lots of new
bugs beside the code duplication.
The third option is to drop some HP-UX signals and shuffle some other signals
around to bring SIGRTMIN to 32. This is of course an ABI change, but testing
has shown that existing Linux installations are not visibly affected by this
change - most likely because we move those signals around which are rarely used
and move them to slots which haven't been used in Linux yet. In an existing
installation I was able to exchange either the Linux kernel or glibc (or both)
without affecting the boot process and installed applications.
Dropping the HP-UX signals isn't an issue either, since support for HP-UX was
basically dropped a few months back with Kernel 3.14 in commit
f5a408d53edef3af07ac7697b8bc54a755628450 already, when we changed EWOULDBLOCK
to be equal to EAGAIN.
So, even if this is an ABI change, it's better to change it now and thus bring
PARISC Linux in sync with other architectures to avoid other issues in the
future.
Signed-off-by: Helge Deller <deller@gmx.de>
Cc: Carlos O'Donell <carlos@systemhalted.org>
Cc: John David Anglin <dave.anglin@bell.net>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Aaro Koskinen <aaro.koskinen@iki.fi>
Cc: PARISC Linux Kernel Mailinglist <linux-parisc@vger.kernel.org>
Tested-by: Aaro Koskinen <aaro.koskinen@iki.fi>
diff --git a/arch/parisc/include/uapi/asm/signal.h b/arch/parisc/include/uapi/asm/signal.h
index f5645d6..10df707 100644
--- a/arch/parisc/include/uapi/asm/signal.h
+++ b/arch/parisc/include/uapi/asm/signal.h
@@ -8,12 +8,12 @@
#define SIGTRAP 5
#define SIGABRT 6
#define SIGIOT 6
-#define SIGEMT 7
+#define SIGSTKFLT 7
#define SIGFPE 8
#define SIGKILL 9
#define SIGBUS 10
#define SIGSEGV 11
-#define SIGSYS 12 /* Linux doesn't use this */
+#define SIGXCPU 12
#define SIGPIPE 13
#define SIGALRM 14
#define SIGTERM 15
@@ -32,16 +32,12 @@
#define SIGTTIN 27
#define SIGTTOU 28
#define SIGURG 29
-#define SIGLOST 30 /* Linux doesn't use this either */
-#define SIGUNUSED 31
-#define SIGRESERVE SIGUNUSED
-
-#define SIGXCPU 33
-#define SIGXFSZ 34
-#define SIGSTKFLT 36
+#define SIGXFSZ 30
+#define SIGUNUSED 31
+#define SIGSYS 31 /* Linux doesn't use this */
/* These should not be considered constants from userland. */
-#define SIGRTMIN 37
+#define SIGRTMIN 32
#define SIGRTMAX _NSIG /* it's 44 under HP/UX */
/*
Reply to: