commit 04b8eb7a4ccd ("symbol lookup: introduce dereference_symbol_descriptor()}"
deprecated vsprintf extension %pf and %pF.
There are approximately these total uses of the symbolic
lookup vsprintf extensions %p[SsFf]:
$ git grep '".*[^%]%p[SsFf]' | \
grep -oh '%p[SsFf]' | sort | uniq -c | sort -rn
231 %pS
65 %ps
60 %pf
47 %pF
which is about a 3:1 ratio favoring %pS
so a script to convert all the %pf uses to %ps and %pF uses to %pS
could be useful.
There are a few files that appear should not be converted.
$ git grep -w --name-only -i '%pf'| \
grep -vP '^(?:Documentation|tools|include/linux/freezer.h)'| \
xargs sed -i -e 's/%pf/%ps/g' -e 's/%pF/%pS/g'
If that script above is run, it leaves the following patch
to be applied:
---
Documentation/core-api/printk-formats.rst | 10 ----------
lib/vsprintf.c | 12 ++----------
2 files changed, 2 insertions(+), 20 deletions(-)
diff --git a/Documentation/core-api/printk-formats.rst b/Documentation/core-api/printk-formats.rst
index ff48b55040ef..ab5a6d1b05c0 100644
--- a/Documentation/core-api/printk-formats.rst
+++ b/Documentation/core-api/printk-formats.rst
@@ -70,8 +70,6 @@ Symbols/Function Pointers
%pS versatile_init+0x0/0x110
%ps versatile_init
- %pF versatile_init+0x0/0x110
- %pf versatile_init
%pSR versatile_init+0x9/0x110
(with __builtin_extract_return_addr() translation)
%pB prev_fn_of_versatile_init+0x88/0x88
@@ -81,14 +79,6 @@ The ``S`` and ``s`` specifiers are used for printing a pointer in symbolic
format. They result in the symbol name with (S) or without (s)
offsets. If KALLSYMS are disabled then the symbol address is printed instead.
-Note, that the ``F`` and ``f`` specifiers are identical to ``S`` (``s``)
-and thus deprecated. We have ``F`` and ``f`` because on ia64, ppc64 and
-parisc64 function pointers are indirect and, in fact, are function
-descriptors, which require additional dereferencing before we can lookup
-the symbol. As of now, ``S`` and ``s`` perform dereferencing on those
-platforms (when needed), so ``F`` and ``f`` exist for compatibility
-reasons only.
-
The ``B`` specifier results in the symbol name with offsets and should be
used when printing stack backtraces. The specifier takes into
consideration the effect of compiler optimisations which may occur
diff --git a/lib/vsprintf.c b/lib/vsprintf.c
index 37a54a6dd594..393002bf5298 100644
--- a/lib/vsprintf.c
+++ b/lib/vsprintf.c
@@ -795,7 +795,7 @@ char *symbol_string(char *buf, char *end, void *ptr,
#ifdef CONFIG_KALLSYMS
if (*fmt == 'B')
sprint_backtrace(sym, value);
- else if (*fmt != 'f' && *fmt != 's')
+ else if (*fmt != 's')
sprint_symbol(sym, value);
else
sprint_symbol_no_offset(sym, value);
@@ -1756,9 +1756,7 @@ char *device_node_string(char *buf, char *end, struct device_node *dn,
*
* - 'S' For symbolic direct pointers (or function descriptors) with offset
* - 's' For symbolic direct pointers (or function descriptors) without offset
- * - 'F' Same as 'S'
- * - 'f' Same as 's'
- * - '[FfSs]R' as above with __builtin_extract_return_addr() translation
+ * - '[Ss]R' as above with __builtin_extract_return_addr() translation
* - 'B' For backtraced symbolic direct pointers with offset
* - 'R' For decoded struct resource, e.g., [mem 0x0-0x1f 64bit pref]
* - 'r' For raw struct resource, e.g., [mem 0x0-0x1f flags 0x201]
@@ -1872,8 +1870,6 @@ char *pointer(const char *fmt, char *buf, char *end, void *ptr,
}
switch (*fmt) {
- case 'F':
- case 'f':
case 'S':
case 's':
ptr = dereference_symbol_descriptor(ptr);
@@ -2603,8 +2599,6 @@ int vbin_printf(u32 *bin_buf, size_t size, const char *fmt, va_list args)
/* Dereference of functions is still OK */
case 'S':
case 's':
- case 'F':
- case 'f':
case 'x':
case 'K':
save_arg(void *);
@@ -2779,8 +2773,6 @@ int bstr_printf(char *buf, size_t size, const char *fmt, const u32 *bin_buf)
switch (*fmt) {
case 'S':
case 's':
- case 'F':
- case 'f':
case 'x':
case 'K':
process = true;
---
Hi,
Cc-ing Petr and Steven
On (11/25/18 01:13), Joe Perches wrote:
> commit 04b8eb7a4ccd ("symbol lookup: introduce dereference_symbol_descriptor()}"
>
> deprecated vsprintf extension %pf and %pF.
>
> There are approximately these total uses of the symbolic
> lookup vsprintf extensions %p[SsFf]:
>
> $ git grep '".*[^%]%p[SsFf]' | \
> grep -oh '%p[SsFf]' | sort | uniq -c | sort -rn
> 231 %pS
> 65 %ps
> 60 %pf
> 47 %pF
I didn't bother to remove "pf/pF" because I didn't want to count on:
a) everyone running checkpatch.pl
b) everyone testing all printk-s they added
I guess pf/pF still can sneak in.
But I don't have a really strong opinion on this. If general
consensus is that we shall remove deprecated specifiers, then
I'm fine.
> which is about a 3:1 ratio favoring %pS
>
> so a script to convert all the %pf uses to %ps and %pF uses to %pS
> could be useful.
>
> There are a few files that appear should not be converted.
>
> $ git grep -w --name-only -i '%pf'| \
> grep -vP '^(?:Documentation|tools|include/linux/freezer.h)'| \
> xargs sed -i -e 's/%pf/%ps/g' -e 's/%pF/%pS/g'
>
> If that script above is run, it leaves the following patch
> to be applied:
After running this script I still have a bunch of leftovers:
//
// these two are probably not really relevant, but still - %pF/%pf
//
tools/lib/traceevent/event-parse.c: if (asprintf(&format, "%%pf: (NO FORMAT FOUND at %llx)\n", addr) < 0)
tools/lib/traceevent/event-parse.c: if (asprintf(&format, "%s: %s", "%pf", printk->printk) < 0)
arch/um/kernel/sysrq.c: pr_info(" [<%08lx>] %s%pF\n", address, reliable ? "" : "? ",
arch/x86/xen/multicalls.c: printk(KERN_DEBUG " call %2d/%d: op=%lu arg=[%lx] result=%ld\t%pF\n",
kernel/async.c: pr_debug("calling %lli_%pF @ %i\n",
kernel/async.c: pr_debug("initcall %lli_%pF returned 0 after %lld usecs\n",
-ss
And I forgot to actually Cc Petr and Steven...
Top-posting.
-------
Hi,
Cc-ing Petr and Steven
On (11/25/18 01:13), Joe Perches wrote:
> commit 04b8eb7a4ccd ("symbol lookup: introduce dereference_symbol_descriptor()}"
>
> deprecated vsprintf extension %pf and %pF.
>
> There are approximately these total uses of the symbolic
> lookup vsprintf extensions %p[SsFf]:
>
> $ git grep '".*[^%]%p[SsFf]' | \
> grep -oh '%p[SsFf]' | sort | uniq -c | sort -rn
> 231 %pS
> 65 %ps
> 60 %pf
> 47 %pF
I didn't bother to remove "pf/pF" because I didn't want to count on:
a) everyone running checkpatch.pl
b) everyone testing all printk-s they added
I guess pf/pF still can sneak in.
But I don't have a really strong opinion on this. If general
consensus is that we shall remove deprecated specifiers, then
I'm fine.
> which is about a 3:1 ratio favoring %pS
>
> so a script to convert all the %pf uses to %ps and %pF uses to %pS
> could be useful.
>
> There are a few files that appear should not be converted.
>
> $ git grep -w --name-only -i '%pf'| \
> grep -vP '^(?:Documentation|tools|include/linux/freezer.h)'| \
> xargs sed -i -e 's/%pf/%ps/g' -e 's/%pF/%pS/g'
>
> If that script above is run, it leaves the following patch
> to be applied:
After running this script I still have a bunch of leftovers:
//
// these two are probably not really relevant, but still - %pF/%pf
//
tools/lib/traceevent/event-parse.c: if (asprintf(&format, "%%pf: (NO FORMAT FOUND at %llx)\n", addr) < 0)
tools/lib/traceevent/event-parse.c: if (asprintf(&format, "%s: %s", "%pf", printk->printk) < 0)
arch/um/kernel/sysrq.c: pr_info(" [<%08lx>] %s%pF\n", address, reliable ? "" : "? ",
arch/x86/xen/multicalls.c: printk(KERN_DEBUG " call %2d/%d: op=%lu arg=[%lx] result=%ld\t%pF\n",
kernel/async.c: pr_debug("calling %lli_%pF @ %i\n",
kernel/async.c: pr_debug("initcall %lli_%pF returned 0 after %lld usecs\n",
-ss
On Mon, 2018-11-26 at 15:08 +0900, Sergey Senozhatsky wrote:
> > There are approximately these total uses of the symbolic
> > lookup vsprintf extensions %p[SsFf]:
> >
> > $ git grep '".*[^%]%p[SsFf]' | \
> > grep -oh '%p[SsFf]' | sort | uniq -c | sort -rn
> > 231 %pS
> > 65 %ps
> > 60 %pf
> > 47 %pF
>
> I didn't bother to remove "pf/pF" because I didn't want to count on:
> a) everyone running checkpatch.pl
> b) everyone testing all printk-s they added
>
> I guess pf/pF still can sneak in.
Any use of %p<invalid_extension> could sneak in.
And thanks, the checkpatch test for %p<invalid_extensions>
should be updated too.
> But I don't have a really strong opinion on this. If general
> consensus is that we shall remove deprecated specifiers, then
> I'm fine.
I think converting deprecated things is generally good.
Also, the %p<extension> lettering space is limited, so
cleaning deprecated extensions is useful.
> After running this script I still have a bunch of leftovers:
>
> //
> // these two are probably not really relevant, but still - %pF/%pf
> //
> tools/lib/traceevent/event-parse.c: if (asprintf(&format, "%%pf: (NO FORMAT FOUND at %llx)\n", addr) < 0)
> tools/lib/traceevent/event-parse.c: if (asprintf(&format, "%s: %s", "%pf", printk->printk) < 0)
I was not/am not sure about those.
> arch/um/kernel/sysrq.c: pr_info(" [<%08lx>] %s%pF\n", address, reliable ? "" : "? ",
> arch/x86/xen/multicalls.c: printk(KERN_DEBUG " call %2d/%d: op=%lu arg=[%lx] result=%ld\t%pF\n",
> kernel/async.c: pr_debug("calling %lli_%pF @ %i\n",
> kernel/async.c: pr_debug("initcall %lli_%pF returned 0 after %lld usecs\n",
Missed those by using \b, thanks again.
On Sun 2018-11-25 01:13:51, Joe Perches wrote:
> commit 04b8eb7a4ccd ("symbol lookup: introduce dereference_symbol_descriptor()}"
>
> deprecated vsprintf extension %pf and %pF.
>
> so a script to convert all the %pf uses to %ps and %pF uses to %pS
> could be useful.
>
> There are a few files that appear should not be converted.
>
> $ git grep -w --name-only -i '%pf'| \
> grep -vP '^(?:Documentation|tools|include/linux/freezer.h)'| \
> xargs sed -i -e 's/%pf/%ps/g' -e 's/%pF/%pS/g'
>
> If that script above is run, it leaves the following patch
> to be applied:
> diff --git a/lib/vsprintf.c b/lib/vsprintf.c
> index 37a54a6dd594..393002bf5298 100644
> --- a/lib/vsprintf.c
> +++ b/lib/vsprintf.c
> @@ -1872,8 +1870,6 @@ char *pointer(const char *fmt, char *buf, char *end, void *ptr,
> }
>
> switch (*fmt) {
> - case 'F':
> - case 'f':
Please, keep handling these modifiers so that they do not get reused
anytime soon.
The pointer modifiers are evil. Any misuse can easily lead to a crash.
Any mistakes with upstreaming 3rd party patches or backporting stable
fixes would be hard to notice.
Well, it is perfectly fine to just explicitly fallback to
the default %p behavior. I mean something like:
case 'F':
case 'f':
/* Replaced by %ps and %pS and removed in 4.21 */
break;
Best Regards,
Petr