Skip to contentSkip to conversation

parisc: Use irq_enter_rcu() to fix warning at kernel/context_tracking.c:367

PATCH New [email protected] 1 participant 7 messages Jan 23, 2024 PA-RISC architecture

lore
Share & tools
lore
#0 lore raw

PATCH 1/7 parisc: Use irq_enter_rcu() to fix warning at kernel/context_tracking.c:367

From: Helge Deller <[email protected]>

Use irq*_rcu() functions to fix this kernel warning:

 WARNING: CPU: 0 PID: 0 at kernel/context_tracking.c:367 ct_irq_enter+0xa0/0xd0
 Modules linked in:
 CPU: 0 PID: 0 Comm: swapper/0 Not tainted 6.7.0-rc3-64bit+ #1037
 Hardware name: 9000/785/C3700

IASQ: 0000000000000000 0000000000000000 IAOQ: 00000000412cd758 00000000412cd75c
IIR: 03ffe01f ISR: 0000000000000000 IOR: 0000000043c20c20
CPU: 0 CR30: 0000000041caa000 CR31: 0000000000000000
ORIG_R28: 0000000000000005
IAOQ[0]: ct_irq_enter+0xa0/0xd0
IAOQ[1]: ct_irq_enter+0xa4/0xd0
RP(r2): irq_enter+0x34/0x68

 Backtrace:
  [<000000004034a3ec>] irq_enter+0x34/0x68
  [<000000004030dc48>] do_cpu_irq_mask+0xc0/0x450
  [<0000000040303070>] intr_return+0x0/0xc
---
 arch/parisc/kernel/irq.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/parisc/kernel/irq.c +2 -2
index 2f81bfd4f15e..dff66be65d29 100644
@@ -498,7 +498,7 @@ asmlinkage void do_cpu_irq_mask(struct pt_regs *regs)
 
 	old_regs = set_irq_regs(regs);
 	local_irq_disable();
-	irq_enter();
+	irq_enter_rcu();
 
 	eirr_val = mfctl(23) & cpu_eiem & per_cpu(local_ack_eiem, cpu);
 	if (!eirr_val)
@@ -533,7 +533,7 @@ asmlinkage void do_cpu_irq_mask(struct pt_regs *regs)
 #endif /* CONFIG_IRQSTACKS */
 
  out:
-	irq_exit();
+	irq_exit_rcu();
 	set_irq_regs(old_regs);
 	return;
 
Signature · 1 line
2.43.0
#1 lore raw

PATCH 2/7 parisc: Make RO_DATA page aligned in vmlinux.lds.S

From: Helge Deller <[email protected]>

The rodata_test program for CONFIG_DEBUG_RODATA_TEST=y complains if read-only data does not start at page boundary.

---
 arch/parisc/kernel/vmlinux.lds.S | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/parisc/kernel/vmlinux.lds.S +1 -1
index 548051b0b4af..b445e47903cf 100644
@@ -127,7 +127,7 @@ SECTIONS
 	}
 #endif
 
-	RO_DATA(8)
+	RO_DATA(PAGE_SIZE)
 
 	/* unwind info */
 	. = ALIGN(4);
Signature · 1 line
2.43.0
#2 lore raw

PATCH 3/7 parisc: Check for valid stride size for cache flushes

In reply to [email protected]

From: Helge Deller <[email protected]>

Report if the calculated cache stride size is zero, otherwise the cache flushing routine will never finish and hang the machine.
This can be reproduced with a testcase in qemu, where the firmware reports wrong cache values.

---
 arch/parisc/kernel/cache.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/parisc/kernel/cache.c +4 -0
index 268d90a9325b..0c015487e5db 100644
@@ -264,6 +264,10 @@ parisc_cache_init(void)
 	icache_stride = CAFL_STRIDE(cache_info.ic_conf);
 #undef CAFL_STRIDE
 
+	/* stride needs to be non-zero, otherwise cache flushes will not work */
+	WARN_ON(cache_info.dc_size && dcache_stride == 0);
+	WARN_ON(cache_info.ic_size && icache_stride == 0);
+
 	if ((boot_cpu_data.pdc.capabilities & PDC_MODEL_NVA_MASK) ==
 						PDC_MODEL_NVA_UNSUPPORTED) {
 		printk(KERN_WARNING "parisc_cache_init: Only equivalent aliasing supported!\n");
Signature · 1 line
2.43.0
#3 lore raw

PATCH 4/7 parisc: Prevent hung tasks when printing inventory on serial console

In reply to [email protected]

From: Helge Deller <[email protected]>

Printing the inventory on a serial console can be quite slow and thus may trigger the hung task detector (CONFIG_DETECT_HUNG_TASK=y) and possibly reboot the machine. Adding a cond_resched() prevents this.

---
 arch/parisc/kernel/drivers.c | 3 +++
 1 file changed, 3 insertions(+)
arch/parisc/kernel/drivers.c +3 -0
index 25f9b9e9d6df..404ea37705ce 100644
@@ -1004,6 +1004,9 @@ static __init int qemu_print_iodc_data(struct device *lin_dev, void *data)
 
 	pr_info("\n");
 
+	/* Prevent hung task messages when printing on serial console */
+	cond_resched();
+
 	pr_info("#define HPA_%08lx_DESCRIPTION \"%s\"\n",
 		hpa, parisc_hardware_description(&dev->id));
 
Signature · 1 line
2.43.0
#4 lore raw

PATCH 5/7 parisc: Drop unneeded semicolon in parse_tree_node()

In reply to [email protected]

From: Helge Deller <[email protected]>

---
 arch/parisc/kernel/drivers.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/parisc/kernel/drivers.c +1 -1
index 404ea37705ce..c7ff339732ba 100644
@@ -742,7 +742,7 @@ parse_tree_node(struct device *parent, int index, struct hardware_path *modpath)
 	};
 
 	if (device_for_each_child(parent, &recurse_data, descend_children))
-		{ /* nothing */ };
+		{ /* nothing */ }
 
 	return d.dev;
 }
Signature · 1 line
2.43.0
#5 lore raw

PATCH 6/7 parisc: Fix random data corruption from exception handler

In reply to [email protected]

From: Helge Deller <[email protected]>

The current exception handler implementation, which assists when accessing user space memory, may exhibit random data corruption if the compiler decides to use a different register than the given register %r29 (defined in ASM_EXCEPTIONTABLE_REG) for the error code. If the compiler choose another register, the fault handler will nevertheless store -EFAULT into %r29 and thus trash whatever this register is used for.
Looking at the assembly I found that this happened e.g. in emulate_ldd().

To solve the issue, the easiest solution would be if it somehow is possible to tell the fault handler which register is used to hold the error code. Using %0 or %1 in the inline assembly is not posssible as it will show up as e.g. %r29 (with the "%r" prefix), which the GNU assembler can not convert to an integer.

This patch takes another, better and more flexible approach:
We extend the __ex_table (which is out of the execution path) by one 32-word. In this word we tell the compiler to insert the assembler instruction "ldi 1,%reg", where %reg references the register which the compiler choosed for the error return code.
In case of an access failure, the fault handler finds the __ex_table entry and can examine the opcode. The used register is encoded in the lowest 5 bits, and the fault handler can then store -EFAULT into this register.

---
 arch/parisc/include/asm/assembly.h      |  1 +
 arch/parisc/include/asm/extable.h       | 64 +++++++++++++++++++++++++
 arch/parisc/include/asm/special_insns.h |  4 +-
 arch/parisc/include/asm/uaccess.h       | 48 +++----------------
 arch/parisc/kernel/cache.c              |  4 +-
 arch/parisc/kernel/unaligned.c          | 44 ++++++++---------
 arch/parisc/mm/fault.c                  | 10 ++--
 7 files changed, 105 insertions(+), 70 deletions(-)
 create mode 100644 arch/parisc/include/asm/extable.h
arch/parisc/include/asm/assembly.h +1 -0
index 74d17d7e759d..6b1aaef07f52 100644
@@ -576,6 +576,7 @@
 	.section __ex_table,"aw"			!	\
 	.align 4					!	\
 	.word (fault_addr - .), (except_addr - .)	!	\
+	ldi 0,%r0					!	\
 	.previous
 
 
arch/parisc/include/asm/extable.h new file +64 -0

64 lines not on this page · raw message

arch/parisc/include/asm/special_insns.h +2 -2
index c822bd0c0e3c..3c07d1f4893c 100644
@@ -8,7 +8,7 @@
 		"copy %%r0,%0\n"			\
 		"8:\tlpa %%r0(%1),%0\n"			\
 		"9:\n"					\
-		ASM_EXCEPTIONTABLE_ENTRY(8b, 9b)	\
+		ASM_EXCEPTIONTABLE_ENTRY(8b, 9b, "nop")	\
 		: "=&r" (pa)				\
 		: "r" (va)				\
 		: "memory"				\
@@ -22,7 +22,7 @@
 		"copy %%r0,%0\n"			\
 		"8:\tlpa %%r0(%%sr3,%1),%0\n"		\
 		"9:\n"					\
-		ASM_EXCEPTIONTABLE_ENTRY(8b, 9b)	\
+		ASM_EXCEPTIONTABLE_ENTRY(8b, 9b, "nop")	\
 		: "=&r" (pa)				\
 		: "r" (va)				\
 		: "memory"				\
arch/parisc/include/asm/uaccess.h +7 -41

94 lines not on this page · raw message

arch/parisc/kernel/cache.c +2 -2
index 0c015487e5db..5552602fcaef 100644
@@ -854,7 +854,7 @@ SYSCALL_DEFINE3(cacheflush, unsigned long, addr, unsigned long, bytes,
 #endif
 			"   fic,m	%3(%4,%0)\n"
 			"2: sync\n"
-			ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 2b)
+			ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 2b, "%1")
 			: "+r" (start), "+r" (error)
 			: "r" (end), "r" (dcache_stride), "i" (SR_USER));
 	}
@@ -869,7 +869,7 @@ SYSCALL_DEFINE3(cacheflush, unsigned long, addr, unsigned long, bytes,
 #endif
 			"   fdc,m	%3(%4,%0)\n"
 			"2: sync\n"
-			ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 2b)
+			ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 2b, "%1")
 			: "+r" (start), "+r" (error)
 			: "r" (end), "r" (icache_stride), "i" (SR_USER));
 	}
arch/parisc/kernel/unaligned.c +22 -22

92 lines not on this page · raw message

arch/parisc/mm/fault.c +7 -3
index 2fe5b44986e0..c303b6f90e0e 100644
@@ -150,11 +150,15 @@ int fixup_exception(struct pt_regs *regs)
 		 * Fix up get_user() and put_user().
 		 * ASM_EXCEPTIONTABLE_ENTRY_EFAULT() sets the least-significant
 		 * bit in the relative address of the fixup routine to indicate
-		 * that gr[ASM_EXCEPTIONTABLE_REG] should be loaded with
-		 * -EFAULT to report a userspace access error.
+		 * that the register, encoded in the additonal "ldi" opcode, should
+		 * be loaded with -EFAULT to report a userspace access error.
 		 */
 		if (fix->fixup & 1) {
-			regs->gr[ASM_EXCEPTIONTABLE_REG] = -EFAULT;
+			int fault_error_reg = fix->err_opcode & 0x1f;
+			if (fault_error_reg)
+				regs->gr[fault_error_reg] = -EFAULT;
+			pr_debug("FIXUP REG %d at %pS\n", fault_error_reg,
+				(void*)regs->iaoq[0]);
 
 			/* zero target register for get_user() */
 			if (parisc_acctyp(0, regs->iir) == VM_READ) {
Signature · 1 line
2.43.0
#6 lore raw

PATCH 7/7 parisc: Show kernel unaligned memory accesses

In reply to [email protected]

From: Helge Deller <[email protected]>

Warn if some kernel function triggers unaligned memory accesses.

---
 arch/parisc/kernel/unaligned.c | 7 +++++++
 1 file changed, 7 insertions(+)
arch/parisc/kernel/unaligned.c +7 -0
index c520e551a165..95a2741f26f3 100644
@@ -399,6 +399,13 @@ void handle_unaligned(struct pt_regs *regs)
 
 		if (!unaligned_enabled)
 			goto force_sigbus;
+	} else {
+		static DEFINE_RATELIMIT_STATE(kernel_ratelimit, 5 * HZ, 5);
+		if (!(current->thread.flags & PARISC_UAC_NOPRINT) &&
+			__ratelimit(&kernel_ratelimit))
+			pr_warn("Kernel: unaligned access to " RFMT " in %pS "
+					"(iir " RFMT ")\n",
+				regs->ior, (void *)regs->iaoq[0], regs->iir);
 	}
 
 	/* handle modification - OK, it's ugly, see the instruction manual */
Signature · 1 line
2.43.0