Skip to contentSkip to conversation

libfdt: Fix signed/unsigned comparison warnings

PATCH Applied Andre Przywara 2 participants 37 messages Sep 21, 2020 – Oct 2, 2020 22 messages not in the archive

v1 of 2 Applied Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover libfdt: Fix signed/unsigned comparison warnings

When libfdt is compiled with -Wsign-compare or -Wextra, GCC emits quite
some warnings about the signedness of the operands not matching:
=================
libfdt/fdt.c:140:18: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
if ((absoffset < offset)
.....
=================

This does not occur under normal conditions in the dtc repo, but might show up when libfdt is embedded in another project. There have been reports from U-Boot and Trusted-Firmware-A.

The underlying issue is mostly due to C's promotion behaviour (ANSI C section 6.1.3.8) when dealing with operands of different signedness (but same size): Signed values get implictly casted to unsigned, which is not typically what we want if they could have been negative.

The Internet(TM) suggests that blindly applying casts is probably doing more harm than it helps, so this series tries to fix the underlying issues properly.
In libfdt, some types are somewhat suboptimal ("int bufsize" comes to mind); some signed types are due to them being returned along wih error values in other functions (node offsets).
So these fixes here have been based on the following assumptions:
- We cannot change the prototype of exported functions.
- It's better to change types (for local variables) than to cast.
- If we have established that a signed value is not negative, we can safely cast it to an unsigned type.

I split up the fixes in small chunks, to make them easier to review. The first four patches change types, the next six ones use casts after we made sure the values are not negative.

This is only covering libfdt for now (which is what those other projects care about). There are more issues with dtc, but they can be addressed later.

Not sure if some of these checks should be gated by can_assume() calls.

Please have a look, happy to discuss the invididual cases.

Cheers,
Andre

Andre Przywara (13):
libfdt: fdt_offset_ptr(): Fix comparison warnings
libfdt: fdt_mem_rsv(): Fix comparison warnings
libfdt: fdt_grab_space_(): Fix comparison warning
libfdt: fdt_add_string_(): Fix comparison warning
libfdt: fdt_move(): Fix comparison warnings
libfdt: fdt_splice_(): Fix comparison warning
libfdt: fdt_create_with_flags(): Fix comparison warning
libfdt: fdt_resize(): Fix comparison warning
libfdt: libfdt_wip: Fix comparison warning
libfdt: overlay: Fix comparison warning
libfdt: fdt_get_string(): Fix sequential write comparison warnings
libfdt: fdt_node_offset_by_phandle(): Fix comparison warning
libfdt: fdt_strerror(): Fix comparison warning

Simon Glass (1):
libfdt: fdt_get_string(): Fix comparison warnings

 libfdt/fdt.c          | 15 +++++++++++----
 libfdt/fdt_overlay.c  |  3 ++-
 libfdt/fdt_ro.c       | 14 +++++++-------
 libfdt/fdt_rw.c       |  2 +-
 libfdt/fdt_strerror.c |  2 +-
 libfdt/fdt_sw.c       | 14 +++++++++-----
 libfdt/fdt_wip.c      |  4 ++--
 7 files changed, 33 insertions(+), 21 deletions(-)
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-1-andre.przywara-5wv7dgnIgG8@public.gmane.org

#2 lore raw

PATCH 1/14 libfdt: fdt_offset_ptr(): Fix comparison warnings

With -Wsign-compare, compilers warn about mismatching signedness in comparisons in fdt_offset_ptr().

This mostly stems from "offset" being passed in as a signed integer, even though the function would not really tolerate negative values.

Short of changing the prototype, check that offset is not negative, and use an unsigned type internally.

---
 libfdt/fdt.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
libfdt/fdt.c +7 -3
index 37b7b93..04e1e06 100644
@@ -134,16 +134,20 @@ int fdt_check_header(const void *fdt)
 
 const void *fdt_offset_ptr(const void *fdt, int offset, unsigned int len)
 {
-	unsigned absoffset = offset + fdt_off_dt_struct(fdt);
+	unsigned int uoffset = offset;
+	unsigned int absoffset = offset + fdt_off_dt_struct(fdt);
+
+	if (offset < 0)
+		return NULL;
 
 	if (!can_assume(VALID_INPUT))
-		if ((absoffset < offset)
+		if ((absoffset < uoffset)
 		    || ((absoffset + len) < absoffset)
 		    || (absoffset + len) > fdt_totalsize(fdt))
 			return NULL;
 
 	if (can_assume(LATEST) || fdt_version(fdt) >= 0x11)
-		if (((offset + len) < offset)
+		if (((uoffset + len) < uoffset)
 		    || ((offset + len) > fdt_size_dt_struct(fdt)))
 			return NULL;
 
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-2-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#4 lore raw

Re: [PATCH 01/14] libfdt: fdt_offset_ptr(): Fix comparison warnings

Quoted 10 lines · Andre Przywara

With -Wsign-compare, compilers warn about mismatching signedness in comparisons in fdt_offset_ptr().

This mostly stems from "offset" being passed in as a signed integer, even though the function would not really tolerate negative values.

Short of changing the prototype, check that offset is not negative, and use an unsigned type internally.

Applied, thanks.

Quoted 32 lines
---
 libfdt/fdt.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
libfdt/fdt.c +7 -3
index 37b7b93..04e1e06 100644
@@ -134,16 +134,20 @@ int fdt_check_header(const void *fdt)
 
 const void *fdt_offset_ptr(const void *fdt, int offset, unsigned int len)
 {
-	unsigned absoffset = offset + fdt_off_dt_struct(fdt);
+	unsigned int uoffset = offset;
+	unsigned int absoffset = offset + fdt_off_dt_struct(fdt);
+
+	if (offset < 0)
+		return NULL;
 
 	if (!can_assume(VALID_INPUT))
-		if ((absoffset < offset)
+		if ((absoffset < uoffset)
 		    || ((absoffset + len) < absoffset)
 		    || (absoffset + len) > fdt_totalsize(fdt))
 			return NULL;
 
 	if (can_assume(LATEST) || fdt_version(fdt) >= 0x11)
-		if (((offset + len) < offset)
+		if (((uoffset + len) < uoffset)
 		    || ((offset + len) > fdt_size_dt_struct(fdt)))
 			return NULL;
 
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#5 lore raw

PATCH 2/14 libfdt: fdt_mem_rsv(): Fix comparison warnings

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_mem_rsv().

Since all involved values must be positive, change the used types to be unsigned.

---
 libfdt/fdt_ro.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_ro.c +2 -2
index e03570a..3c8d143 100644
@@ -157,8 +157,8 @@ int fdt_generate_phandle(const void *fdt, uint32_t *phandle)
 
 static const struct fdt_reserve_entry *fdt_mem_rsv(const void *fdt, int n)
 {
-	int offset = n * sizeof(struct fdt_reserve_entry);
-	int absoffset = fdt_off_mem_rsvmap(fdt) + offset;
+	unsigned int offset = n * sizeof(struct fdt_reserve_entry);
+	unsigned int absoffset = fdt_off_mem_rsvmap(fdt) + offset;
 
 	if (!can_assume(VALID_INPUT)) {
 		if (absoffset < fdt_off_mem_rsvmap(fdt))
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-3-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#7 lore raw

Re: [PATCH 02/14] libfdt: fdt_mem_rsv(): Fix comparison warnings

Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_mem_rsv().

Since all involved values must be positive, change the used types to be unsigned.

Applied, thanks.

Quoted 19 lines
---
 libfdt/fdt_ro.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_ro.c +2 -2
index e03570a..3c8d143 100644
@@ -157,8 +157,8 @@ int fdt_generate_phandle(const void *fdt, uint32_t *phandle)
 
 static const struct fdt_reserve_entry *fdt_mem_rsv(const void *fdt, int n)
 {
-	int offset = n * sizeof(struct fdt_reserve_entry);
-	int absoffset = fdt_off_mem_rsvmap(fdt) + offset;
+	unsigned int offset = n * sizeof(struct fdt_reserve_entry);
+	unsigned int absoffset = fdt_off_mem_rsvmap(fdt) + offset;
 
 	if (!can_assume(VALID_INPUT)) {
 		if (absoffset < fdt_off_mem_rsvmap(fdt))
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#8 lore raw

PATCH 3/14 libfdt: fdt_grab_space_(): Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_grab_space_().

All the involved values cannot be negative, so let's switch the types of the local variables to unsigned to make the compiler happy.

---
 libfdt/fdt_sw.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_sw.c +2 -2
index 94ce4bb..d10a720 100644
@@ -93,8 +93,8 @@ static inline uint32_t sw_flags(void *fdt)
 
 static void *fdt_grab_space_(void *fdt, size_t len)
 {
-	int offset = fdt_size_dt_struct(fdt);
-	int spaceleft;
+	unsigned int offset = fdt_size_dt_struct(fdt);
+	unsigned int spaceleft;
 
 	spaceleft = fdt_totalsize(fdt) - fdt_off_dt_struct(fdt)
 		- fdt_size_dt_strings(fdt);
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-4-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#10 lore raw

Re: [PATCH 03/14] libfdt: fdt_grab_space_(): Fix comparison warning

Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_grab_space_().

All the involved values cannot be negative, so let's switch the types of the local variables to unsigned to make the compiler happy.

Applied, thanks.

Quoted 19 lines
---
 libfdt/fdt_sw.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_sw.c +2 -2
index 94ce4bb..d10a720 100644
@@ -93,8 +93,8 @@ static inline uint32_t sw_flags(void *fdt)
 
 static void *fdt_grab_space_(void *fdt, size_t len)
 {
-	int offset = fdt_size_dt_struct(fdt);
-	int spaceleft;
+	unsigned int offset = fdt_size_dt_struct(fdt);
+	unsigned int spaceleft;
 
 	spaceleft = fdt_totalsize(fdt) - fdt_off_dt_struct(fdt)
 		- fdt_size_dt_strings(fdt);
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#11 lore raw

PATCH 4/14 libfdt: fdt_add_string_(): Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_add_string_().

As struct_top can only be positive, just use an unsigned type for it, and avoid the signedness difference.

---
 libfdt/fdt_sw.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +2 -1
index d10a720..d65e9c8 100644
@@ -249,7 +249,8 @@ static int fdt_add_string_(void *fdt, const char *s)
 	char *strtab = (char *)fdt + fdt_totalsize(fdt);
 	int strtabsize = fdt_size_dt_strings(fdt);
 	int len = strlen(s) + 1;
-	int struct_top, offset;
+	unsigned int struct_top;
+	int offset;
 
 	offset = -strtabsize - len;
 	struct_top = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-5-andre.przywara-5wv7dgnIgG8@public.gmane.org

#13 lore raw

Re: [PATCH 04/14] libfdt: fdt_add_string_(): Fix comparison warning

Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_add_string_().

As struct_top can only be positive, just use an unsigned type for it, and avoid the signedness difference.

I'm not sure this is right. Well.. I'm also not sure it was right before. Adding some more context to explain why..

Quoted 18 lines
---
 libfdt/fdt_sw.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +2 -1
index d10a720..d65e9c8 100644
@@ -249,7 +249,8 @@ static int fdt_add_string_(void *fdt, const char *s)
 	char *strtab = (char *)fdt + fdt_totalsize(fdt);
 	int strtabsize = fdt_size_dt_strings(fdt);
 	int len = strlen(s) + 1;
-	int struct_top, offset;
+	unsigned int struct_top;
+	int offset;
 
 	offset = -strtabsize - len;
 	struct_top = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
Quoted 2 lines
	if (fdt_totalsize(fdt) + offset < struct_top)
		return 0; /* no more room :( */

So strtabsize and len will always be positive (or, if they're not, that's another problem), so offset is always negative. Which means we need the signed addition between totalsize and offset for this to be correct.

So I suspect we want to make 'len' and 'offset' unsigned as well, reverse the sign on offset and make it a subtraction in the if instead of an addition-of-negative.

We might then need to explicitly check for offset < totalsize as well, to cover the overflow case.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

A message that is not in the archive (sent off-list or not yet received): [email protected]

#15 lore raw

Re: [PATCH 04/14] libfdt: fdt_add_string_(): Fix comparison warning

Quoted 38 lines · David Gibson
Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_add_string_().

As struct_top can only be positive, just use an unsigned type for it, and avoid the signedness difference.

I'm not sure this is right. Well.. I'm also not sure it was right before. Adding some more context to explain why..

Quoted 18 lines
---
 libfdt/fdt_sw.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +2 -1
index d10a720..d65e9c8 100644
@@ -249,7 +249,8 @@ static int fdt_add_string_(void *fdt, const char *s)
 	char *strtab = (char *)fdt + fdt_totalsize(fdt);
 	int strtabsize = fdt_size_dt_strings(fdt);
 	int len = strlen(s) + 1;
-	int struct_top, offset;
+	unsigned int struct_top;
+	int offset;
 
 	offset = -strtabsize - len;
 	struct_top = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
Quoted 2 lines
	if (fdt_totalsize(fdt) + offset < struct_top)
		return 0; /* no more room :( */

So strtabsize and len will always be positive (or, if they're not, that's another problem), so offset is always negative. Which means we need the signed addition between totalsize and offset for this to be correct.

Quoted 3 lines

So I suspect we want to make 'len' and 'offset' unsigned as well, reverse the sign on offset and make it a subtraction in the if instead of an addition-of-negative.

Ah, yes, much better! To be honest I just did my due diligence on *this* function before, because all those minus signs confused me quite a lot. But your approach makes it much clearer what is going on here.

So if I get this correctly, this function really returns the *negative* offset of the new string, which will end up in the property, for now? And at the end we translate all negative values into the actual offsets? And there is initially a gap between the dt_struct part and the string table, to allow both growing towards each other, without needing to rewrite everything, every time?
Is this documented somewhere? Shall this approach be described in a comment at the top of this file?

Thanks!
Andre.

Quoted 4 lines

We might then need to explicitly check for offset < totalsize as well, to cover the overflow case.

A message that is not in the archive (sent off-list or not yet received): 74c57a12-db86-ec9f-5c91-f0419c24de4b-5wv7dgnIgG8@public.gmane.org

#17 lore raw

Re: [PATCH 04/14] libfdt: fdt_add_string_(): Fix comparison warning

Quoted 47 lines · André Przywara
Quoted 38 lines · David Gibson
Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_add_string_().

As struct_top can only be positive, just use an unsigned type for it, and avoid the signedness difference.

I'm not sure this is right. Well.. I'm also not sure it was right before. Adding some more context to explain why..

Quoted 18 lines
---
 libfdt/fdt_sw.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +2 -1
index d10a720..d65e9c8 100644
@@ -249,7 +249,8 @@ static int fdt_add_string_(void *fdt, const char *s)
 	char *strtab = (char *)fdt + fdt_totalsize(fdt);
 	int strtabsize = fdt_size_dt_strings(fdt);
 	int len = strlen(s) + 1;
-	int struct_top, offset;
+	unsigned int struct_top;
+	int offset;
 
 	offset = -strtabsize - len;
 	struct_top = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
Quoted 2 lines
	if (fdt_totalsize(fdt) + offset < struct_top)
		return 0; /* no more room :( */

So strtabsize and len will always be positive (or, if they're not, that's another problem), so offset is always negative. Which means we need the signed addition between totalsize and offset for this to be correct.

Quoted 3 lines

So I suspect we want to make 'len' and 'offset' unsigned as well, reverse the sign on offset and make it a subtraction in the if instead of an addition-of-negative.

Ah, yes, much better! To be honest I just did my due diligence on *this* function before, because all those minus signs confused me quite a lot. But your approach makes it much clearer what is going on here.

Quoted 6 lines

So if I get this correctly, this function really returns the *negative* offset of the new string, which will end up in the property, for now? And at the end we translate all negative values into the actual offsets? And there is initially a gap between the dt_struct part and the string table, to allow both growing towards each other, without needing to rewrite everything, every time?

Yes, that's correct. During sequential write mode, to avoid excessive memmove()s, the structure block grows up from the bottom of the buffer, while the strings block grows down from the top. Because the "start" (lowest offset) of the strings block is moving, we can't use offsets from that (without having to adjust all the offsets in the structure block all the time). So, we use negative offsets from the "end" (highest offset). I'm scare quoting "start" and "end" because the header records the highest offset of the string block as its "start", which is why the offsets are negative - that makes some of the code paths more common with the normal dtb case.

Quoted 2 lines

Is this documented somewhere? Shall this approach be described in a comment at the top of this file?

No, I don't think it is. So, yes there probably should be a big explanatory comment. Patches welcome?

Note that this isn't supposed to be an externally visible format, you're always supposed to use fdt_finish() to convert it into a regular dtb before handing it off to any other piece of code. We use a different magic number in the header while we're in this state to avoid anything mistaking this for a regular dtb.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

A message that is not in the archive (sent off-list or not yet received): [email protected]

#19 lore raw

Re: [PATCH 04/14] libfdt: fdt_add_string_(): Fix comparison warning

On 02/10/2020 01:01, David Gibson wrote:

Hi David,

Quoted 66 lines · André Przywara
Quoted 47 lines · David Gibson
Quoted 38 lines · Andre Przywara
Quoted 7 lines

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_add_string_().

As struct_top can only be positive, just use an unsigned type for it, and avoid the signedness difference.

I'm not sure this is right. Well.. I'm also not sure it was right before. Adding some more context to explain why..

Quoted 18 lines
---
 libfdt/fdt_sw.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +2 -1
index d10a720..d65e9c8 100644
@@ -249,7 +249,8 @@ static int fdt_add_string_(void *fdt, const char *s)
 	char *strtab = (char *)fdt + fdt_totalsize(fdt);
 	int strtabsize = fdt_size_dt_strings(fdt);
 	int len = strlen(s) + 1;
-	int struct_top, offset;
+	unsigned int struct_top;
+	int offset;
 
 	offset = -strtabsize - len;
 	struct_top = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
Quoted 2 lines
	if (fdt_totalsize(fdt) + offset < struct_top)
		return 0; /* no more room :( */

So strtabsize and len will always be positive (or, if they're not, that's another problem), so offset is always negative. Which means we need the signed addition between totalsize and offset for this to be correct.

Quoted 3 lines

So I suspect we want to make 'len' and 'offset' unsigned as well, reverse the sign on offset and make it a subtraction in the if instead of an addition-of-negative.

Ah, yes, much better! To be honest I just did my due diligence on *this* function before, because all those minus signs confused me quite a lot. But your approach makes it much clearer what is going on here.

Quoted 6 lines

So if I get this correctly, this function really returns the *negative* offset of the new string, which will end up in the property, for now? And at the end we translate all negative values into the actual offsets? And there is initially a gap between the dt_struct part and the string table, to allow both growing towards each other, without needing to rewrite everything, every time?

Yes, that's correct. During sequential write mode, to avoid excessive memmove()s, the structure block grows up from the bottom of the buffer, while the strings block grows down from the top. Because the "start" (lowest offset) of the strings block is moving, we can't use offsets from that (without having to adjust all the offsets in the structure block all the time). So, we use negative offsets from the "end" (highest offset). I'm scare quoting "start" and "end" because the header records the highest offset of the string block as its "start", which is why the offsets are negative - that makes some of the code paths more common with the normal dtb case.

Quoted 5 lines
Quoted 2 lines

Is this documented somewhere? Shall this approach be described in a comment at the top of this file?

No, I don't think it is. So, yes there probably should be a big explanatory comment. Patches welcome?

Yes, that was the idea, just want to double check that I got this right.

Quoted 5 lines

Note that this isn't supposed to be an externally visible format, you're always supposed to use fdt_finish() to convert it into a regular dtb before handing it off to any other piece of code. We use a different magic number in the header while we're in this state to avoid anything mistaking this for a regular dtb.

Yeah, I understand, but the different magics and the negative offsets were quite confusing at first, until I discovered the whole story. Since the library is typically embedded into other projects, I think it's a good idea to document this, in case people follow some rabbit with ctags ...

Cheers,
Andre

A message that is not in the archive (sent off-list or not yet received): 3d6281e8-bd95-b2db-2b80-446f9a98ea66-5wv7dgnIgG8@public.gmane.org

#21 lore raw

Re: [PATCH 04/14] libfdt: fdt_add_string_(): Fix comparison warning

Quoted 83 lines · André Przywara

On 02/10/2020 01:01, David Gibson wrote:

Hi David,

Quoted 66 lines · André Przywara
Quoted 47 lines · David Gibson
Quoted 38 lines · Andre Przywara
Quoted 7 lines

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_add_string_().

As struct_top can only be positive, just use an unsigned type for it, and avoid the signedness difference.

I'm not sure this is right. Well.. I'm also not sure it was right before. Adding some more context to explain why..

Quoted 18 lines
---
 libfdt/fdt_sw.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +2 -1
index d10a720..d65e9c8 100644
@@ -249,7 +249,8 @@ static int fdt_add_string_(void *fdt, const char *s)
 	char *strtab = (char *)fdt + fdt_totalsize(fdt);
 	int strtabsize = fdt_size_dt_strings(fdt);
 	int len = strlen(s) + 1;
-	int struct_top, offset;
+	unsigned int struct_top;
+	int offset;
 
 	offset = -strtabsize - len;
 	struct_top = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
Quoted 2 lines
	if (fdt_totalsize(fdt) + offset < struct_top)
		return 0; /* no more room :( */

So strtabsize and len will always be positive (or, if they're not, that's another problem), so offset is always negative. Which means we need the signed addition between totalsize and offset for this to be correct.

Quoted 3 lines

So I suspect we want to make 'len' and 'offset' unsigned as well, reverse the sign on offset and make it a subtraction in the if instead of an addition-of-negative.

Ah, yes, much better! To be honest I just did my due diligence on *this* function before, because all those minus signs confused me quite a lot. But your approach makes it much clearer what is going on here.

Quoted 6 lines

So if I get this correctly, this function really returns the *negative* offset of the new string, which will end up in the property, for now? And at the end we translate all negative values into the actual offsets? And there is initially a gap between the dt_struct part and the string table, to allow both growing towards each other, without needing to rewrite everything, every time?

Yes, that's correct. During sequential write mode, to avoid excessive memmove()s, the structure block grows up from the bottom of the buffer, while the strings block grows down from the top. Because the "start" (lowest offset) of the strings block is moving, we can't use offsets from that (without having to adjust all the offsets in the structure block all the time). So, we use negative offsets from the "end" (highest offset). I'm scare quoting "start" and "end" because the header records the highest offset of the string block as its "start", which is why the offsets are negative - that makes some of the code paths more common with the normal dtb case.

Quoted 5 lines
Quoted 2 lines

Is this documented somewhere? Shall this approach be described in a comment at the top of this file?

No, I don't think it is. So, yes there probably should be a big explanatory comment. Patches welcome?

Yes, that was the idea, just want to double check that I got this right.

Quoted 5 lines

Note that this isn't supposed to be an externally visible format, you're always supposed to use fdt_finish() to convert it into a regular dtb before handing it off to any other piece of code. We use a different magic number in the header while we're in this state to avoid anything mistaking this for a regular dtb.

Quoted 5 lines

Yeah, I understand, but the different magics and the negative offsets were quite confusing at first, until I discovered the whole story. Since the library is typically embedded into other projects, I think it's a good idea to document this, in case people follow some rabbit with ctags ...

Yes, that makes sense.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#22 lore raw

PATCH 5/14 libfdt: fdt_move(): Fix comparison warnings

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_move().

This stems from "bufsize" being passed in as a signed integer, even though we would expect a buffer size to be positive.

Short of changing the prototype, check that bufsize is not negative, and cast it to an unsigned type in the comparison.

---
 libfdt/fdt.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
libfdt/fdt.c +4 -1
index 04e1e06..43aaada 100644
@@ -314,9 +314,12 @@ const char *fdt_find_string_(const char *strtab, int tabsize, const char *s)
 
 int fdt_move(const void *fdt, void *buf, int bufsize)
 {
+	if (bufsize < 0)
+		return -FDT_ERR_NOSPACE;
+
 	FDT_RO_PROBE(fdt);
 
-	if (fdt_totalsize(fdt) > bufsize)
+	if (fdt_totalsize(fdt) > (unsigned int)bufsize)
 		return -FDT_ERR_NOSPACE;
 
 	memmove(buf, fdt, fdt_totalsize(fdt));
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-6-andre.przywara-5wv7dgnIgG8@public.gmane.org

#24 lore raw

Re: [PATCH 05/14] libfdt: fdt_move(): Fix comparison warnings

Quoted 10 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_move().

This stems from "bufsize" being passed in as a signed integer, even though we would expect a buffer size to be positive.

Short of changing the prototype, check that bufsize is not negative, and cast it to an unsigned type in the comparison.

Logic looks good, but the libfdt-in-tiny-bootloaders people would probably appreciate a !can_assume(VALID_INPUT) test to elide the check on bufsize.

Quoted 22 lines
---
 libfdt/fdt.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
libfdt/fdt.c +4 -1
index 04e1e06..43aaada 100644
@@ -314,9 +314,12 @@ const char *fdt_find_string_(const char *strtab, int tabsize, const char *s)
 
 int fdt_move(const void *fdt, void *buf, int bufsize)
 {
+	if (bufsize < 0)
+		return -FDT_ERR_NOSPACE;
+
 	FDT_RO_PROBE(fdt);
 
-	if (fdt_totalsize(fdt) > bufsize)
+	if (fdt_totalsize(fdt) > (unsigned int)bufsize)
 		return -FDT_ERR_NOSPACE;
 
 	memmove(buf, fdt, fdt_totalsize(fdt));
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#25 lore raw

PATCH 6/14 libfdt: fdt_get_string(): Fix comparison warnings

In reply to a message not in the archive

From: Simon Glass <[email protected]>

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_get_string().

In the first two cases, we have just established that the signed values are not negative, so it's safe to cast the values to an unsigned type.

---
 libfdt/fdt_ro.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_ro.c +2 -2
index 3c8d143..ddb5a2d 100644
@@ -53,7 +53,7 @@ const char *fdt_get_string(const void *fdt, int stroffset, int *lenp)
 
 	err = -FDT_ERR_BADOFFSET;
 	absoffset = stroffset + fdt_off_dt_strings(fdt);
-	if (absoffset >= totalsize)
+	if (absoffset >= (unsigned)totalsize)
 		goto fail;
 	len = totalsize - absoffset;
 
@@ -61,7 +61,7 @@ const char *fdt_get_string(const void *fdt, int stroffset, int *lenp)
 		if (stroffset < 0)
 			goto fail;
 		if (can_assume(LATEST) || fdt_version(fdt) >= 17) {
-			if (stroffset >= fdt_size_dt_strings(fdt))
+			if ((unsigned)stroffset >= fdt_size_dt_strings(fdt))
 				goto fail;
 			if ((fdt_size_dt_strings(fdt) - stroffset) < len)
 				len = fdt_size_dt_strings(fdt) - stroffset;
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-7-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#27 lore raw

Re: [PATCH 06/14] libfdt: fdt_get_string(): Fix comparison warnings

Quoted 10 lines · Andre Przywara

From: Simon Glass <[email protected]>

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_get_string().

In the first two cases, we have just established that the signed values are not negative, so it's safe to cast the values to an unsigned type.

Applied, thanks.

Quoted 26 lines
---
 libfdt/fdt_ro.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_ro.c +2 -2
index 3c8d143..ddb5a2d 100644
@@ -53,7 +53,7 @@ const char *fdt_get_string(const void *fdt, int stroffset, int *lenp)
 
 	err = -FDT_ERR_BADOFFSET;
 	absoffset = stroffset + fdt_off_dt_strings(fdt);
-	if (absoffset >= totalsize)
+	if (absoffset >= (unsigned)totalsize)
 		goto fail;
 	len = totalsize - absoffset;
 
@@ -61,7 +61,7 @@ const char *fdt_get_string(const void *fdt, int stroffset, int *lenp)
 		if (stroffset < 0)
 			goto fail;
 		if (can_assume(LATEST) || fdt_version(fdt) >= 17) {
-			if (stroffset >= fdt_size_dt_strings(fdt))
+			if ((unsigned)stroffset >= fdt_size_dt_strings(fdt))
 				goto fail;
 			if ((fdt_size_dt_strings(fdt) - stroffset) < len)
 				len = fdt_size_dt_strings(fdt) - stroffset;
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#28 lore raw

PATCH 7/14 libfdt: fdt_splice_(): Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_splice_().

Since we just established that oldlen is not negative, we can safely cast it to an unsigned type.

---
 libfdt/fdt_rw.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_rw.c +1 -1
index 93e4a2b..68887b9 100644
@@ -59,7 +59,7 @@ static int fdt_splice_(void *fdt, void *splicepoint, int oldlen, int newlen)
 
 	if ((oldlen < 0) || (soff + oldlen < soff) || (soff + oldlen > dsize))
 		return -FDT_ERR_BADOFFSET;
-	if ((p < (char *)fdt) || (dsize + newlen < oldlen))
+	if ((p < (char *)fdt) || (dsize + newlen < (unsigned)oldlen))
 		return -FDT_ERR_BADOFFSET;
 	if (dsize - oldlen + newlen > fdt_totalsize(fdt))
 		return -FDT_ERR_NOSPACE;
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-8-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#30 lore raw

Re: [PATCH 07/14] libfdt: fdt_splice_(): Fix comparison warning

Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_splice_().

Since we just established that oldlen is not negative, we can safely cast it to an unsigned type.

Applied, thanks.

Quoted 17 lines
---
 libfdt/fdt_rw.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_rw.c +1 -1
index 93e4a2b..68887b9 100644
@@ -59,7 +59,7 @@ static int fdt_splice_(void *fdt, void *splicepoint, int oldlen, int newlen)
 
 	if ((oldlen < 0) || (soff + oldlen < soff) || (soff + oldlen > dsize))
 		return -FDT_ERR_BADOFFSET;
-	if ((p < (char *)fdt) || (dsize + newlen < oldlen))
+	if ((p < (char *)fdt) || (dsize + newlen < (unsigned)oldlen))
 		return -FDT_ERR_BADOFFSET;
 	if (dsize - oldlen + newlen > fdt_totalsize(fdt))
 		return -FDT_ERR_NOSPACE;
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#31 lore raw

PATCH 8/14 libfdt: fdt_create_with_flags(): Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_create_with_flags().

We have just established that bufsize is not negative, so a cast to an unsigned type is safe.

---
 libfdt/fdt_sw.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_sw.c +1 -1
index d65e9c8..8a76adf 100644
@@ -112,7 +112,7 @@ int fdt_create_with_flags(void *buf, int bufsize, uint32_t flags)
 					 sizeof(struct fdt_reserve_entry));
 	void *fdt = buf;
 
-	if (bufsize < hdrsize)
+	if (bufsize < 0 || (unsigned)bufsize < hdrsize)
 		return -FDT_ERR_NOSPACE;
 
 	if (flags & ~FDT_CREATE_FLAGS_ALL)
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-9-andre.przywara-5wv7dgnIgG8@public.gmane.org

#33 lore raw

Re: [PATCH 08/14] libfdt: fdt_create_with_flags(): Fix comparison warning

Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_create_with_flags().

We have just established that bufsize is not negative, so a cast to an unsigned type is safe.

This looks correct, but I think we might be able to do it a bit more neatly by changing the type of hdrsize to signed instead. It's a small value, so we know that's a valid conversion, and that would let your two tests be merged into one here.

Quoted 17 lines
---
 libfdt/fdt_sw.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_sw.c +1 -1
index d65e9c8..8a76adf 100644
@@ -112,7 +112,7 @@ int fdt_create_with_flags(void *buf, int bufsize, uint32_t flags)
 					 sizeof(struct fdt_reserve_entry));
 	void *fdt = buf;
 
-	if (bufsize < hdrsize)
+	if (bufsize < 0 || (unsigned)bufsize < hdrsize)
 		return -FDT_ERR_NOSPACE;
 
 	if (flags & ~FDT_CREATE_FLAGS_ALL)
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#34 lore raw

PATCH 9/14 libfdt: fdt_resize(): Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_resize().

A negative buffer size will surely do us no good, so let's rule this case out first.
In the actual comparison we then know that a cast to an unsigned type is safe.

---
 libfdt/fdt_sw.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +4 -1
index 8a76adf..4c62922 100644
@@ -152,6 +152,9 @@ int fdt_resize(void *fdt, void *buf, int bufsize)
 
 	FDT_SW_PROBE(fdt);
 
+	if (bufsize < 0)
+		return -FDT_ERR_NOSPACE;
+
 	headsize = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
 	tailsize = fdt_size_dt_strings(fdt);
 
@@ -159,7 +162,7 @@ int fdt_resize(void *fdt, void *buf, int bufsize)
 	    headsize + tailsize > fdt_totalsize(fdt))
 		return -FDT_ERR_INTERNAL;
 
-	if ((headsize + tailsize) > bufsize)
+	if ((headsize + tailsize) > (unsigned)bufsize)
 		return -FDT_ERR_NOSPACE;
 
 	oldtail = (char *)fdt + fdt_totalsize(fdt) - tailsize;
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-10-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#36 lore raw

Re: [PATCH 09/14] libfdt: fdt_resize(): Fix comparison warning

Quoted 9 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_resize().

A negative buffer size will surely do us no good, so let's rule this case out first.
In the actual comparison we then know that a cast to an unsigned type is safe.

Applied, thanks.

Quoted 27 lines
---
 libfdt/fdt_sw.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
libfdt/fdt_sw.c +4 -1
index 8a76adf..4c62922 100644
@@ -152,6 +152,9 @@ int fdt_resize(void *fdt, void *buf, int bufsize)
 
 	FDT_SW_PROBE(fdt);
 
+	if (bufsize < 0)
+		return -FDT_ERR_NOSPACE;
+
 	headsize = fdt_off_dt_struct(fdt) + fdt_size_dt_struct(fdt);
 	tailsize = fdt_size_dt_strings(fdt);
 
@@ -159,7 +162,7 @@ int fdt_resize(void *fdt, void *buf, int bufsize)
 	    headsize + tailsize > fdt_totalsize(fdt))
 		return -FDT_ERR_INTERNAL;
 
-	if ((headsize + tailsize) > bufsize)
+	if ((headsize + tailsize) > (unsigned)bufsize)
 		return -FDT_ERR_NOSPACE;
 
 	oldtail = (char *)fdt + fdt_totalsize(fdt) - tailsize;
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#37 lore raw

PATCH 10/14 libfdt: libfdt_wip: Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_setprop_inplace_namelen_partial().

Since fdt_getprop_namelen() can return negative error values, check for those first and bail out early.
Later on we can then safely cast the value to an unsigned type.

---
 libfdt/fdt_wip.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_wip.c +2 -2
index f64139e..d1b4f1b 100644
@@ -20,10 +20,10 @@ int fdt_setprop_inplace_namelen_partial(void *fdt, int nodeoffset,
 
 	propval = fdt_getprop_namelen_w(fdt, nodeoffset, name, namelen,
 					&proplen);
-	if (!propval)
+	if (!propval || proplen < 0)
 		return proplen;
 
-	if (proplen < (len + idx))
+	if ((unsigned)proplen < (len + idx))
 		return -FDT_ERR_NOSPACE;
 
 	memcpy((char *)propval + idx, val, len);
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-11-andre.przywara-5wv7dgnIgG8@public.gmane.org

#39 lore raw

Re: [PATCH 10/14] libfdt: libfdt_wip: Fix comparison warning

Quoted 17 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_setprop_inplace_namelen_partial().

Since fdt_getprop_namelen() can return negative error values, check for those first and bail out early.
Later on we can then safely cast the value to an unsigned type.

---
 libfdt/fdt_wip.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_wip.c +0 -0
index f64139e..d1b4f1b 100644

libfdt/fdt_wip.c int fdt_setprop_inplace_namelen_partial(void *fdt, int nodeoffset, · line 24

Quoted 6 lines
libfdt/fdt_wip.c +1 -1
@@ -20,10 +20,10 @@ int fdt_setprop_inplace_namelen_partial(void *fdt, int nodeoffset,
 
 	propval = fdt_getprop_namelen_w(fdt, nodeoffset, name, namelen,
 					&proplen);
-	if (!propval)
+	if (!propval || proplen < 0)
 		return proplen;

We don't actually need that test - if proplen < 0 with propval non-NULL, that's a bug in fdt_getprop_namelen_w().

Quoted 5 lines
libfdt/fdt_wip.c +1 -1
-	if (proplen < (len + idx))
+	if ((unsigned)proplen < (len + idx))
 		return -FDT_ERR_NOSPACE;
 
 	memcpy((char *)propval + idx, val, len);
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#40 lore raw

PATCH 11/14 libfdt: overlay: Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in overlay_update_local_node_references().

This happens because the division of a signed int by an unsigned int promotes the dividend to unsigned first (ANSI C standard 6.1.3.8).

As in this case we basically just divide by 4, we can do the division separately earlier, which preserves the original type.

---
 libfdt/fdt_overlay.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_overlay.c +2 -1
index b310e49..d217e79 100644
@@ -241,6 +241,7 @@ static int overlay_update_local_node_references(void *fdto,
 
 		if (fixup_len % sizeof(uint32_t))
 			return -FDT_ERR_BADOVERLAY;
+		fixup_len /= sizeof(uint32_t);
 
 		tree_val = fdt_getprop(fdto, tree_node, name, &tree_len);
 		if (!tree_val) {
@@ -250,7 +251,7 @@ static int overlay_update_local_node_references(void *fdto,
 			return tree_len;
 		}
 
-		for (i = 0; i < (fixup_len / sizeof(uint32_t)); i++) {
+		for (i = 0; i < fixup_len; i++) {
 			fdt32_t adj_val;
 			uint32_t poffset;
 
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-12-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#42 lore raw

Re: [PATCH 11/14] libfdt: overlay: Fix comparison warning

Quoted 10 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in overlay_update_local_node_references().

This happens because the division of a signed int by an unsigned int promotes the dividend to unsigned first (ANSI C standard 6.1.3.8).

As in this case we basically just divide by 4, we can do the division separately earlier, which preserves the original type.

Applied, thanks.

Quoted 25 lines
---
 libfdt/fdt_overlay.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
libfdt/fdt_overlay.c +2 -1
index b310e49..d217e79 100644
@@ -241,6 +241,7 @@ static int overlay_update_local_node_references(void *fdto,
 
 		if (fixup_len % sizeof(uint32_t))
 			return -FDT_ERR_BADOVERLAY;
+		fixup_len /= sizeof(uint32_t);
 
 		tree_val = fdt_getprop(fdto, tree_node, name, &tree_len);
 		if (!tree_val) {
@@ -250,7 +251,7 @@ static int overlay_update_local_node_references(void *fdto,
 			return tree_len;
 		}
 
-		for (i = 0; i < (fixup_len / sizeof(uint32_t)); i++) {
+		for (i = 0; i < fixup_len; i++) {
 			fdt32_t adj_val;
 			uint32_t poffset;
 
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#43 lore raw

PATCH 12/14 libfdt: fdt_get_string(): Fix sequential write comparison warnings

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_get_string().

The two occassions dealing with the sequential write case are a bit tricky, since we deliberately abuse the negative values here.

As we have just established that stroffset is negative, we can use casts to appease the compiler.

---
 libfdt/fdt_ro.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_ro.c +2 -2
index ddb5a2d..059d302 100644
@@ -68,9 +68,9 @@ const char *fdt_get_string(const void *fdt, int stroffset, int *lenp)
 		}
 	} else if (fdt_magic(fdt) == FDT_SW_MAGIC) {
 		if ((stroffset >= 0)
-		    || (stroffset < -fdt_size_dt_strings(fdt)))
+		    || (stroffset < -(int)fdt_size_dt_strings(fdt)))
 			goto fail;
-		if ((-stroffset) < len)
+		if ((unsigned)(-stroffset) < len)
 			len = -stroffset;
 	} else {
 		err = -FDT_ERR_INTERNAL;
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-13-andre.przywara-5wv7dgnIgG8@public.gmane.org

#45 lore raw

Re: [PATCH 12/14] libfdt: fdt_get_string(): Fix sequential write comparison warnings

Quoted 19 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_get_string().

The two occassions dealing with the sequential write case are a bit tricky, since we deliberately abuse the negative values here.

As we have just established that stroffset is negative, we can use casts to appease the compiler.

---
 libfdt/fdt_ro.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
libfdt/fdt_ro.c +0 -0
index ddb5a2d..059d302 100644

libfdt/fdt_ro.c const char *fdt_get_string(const void *fdt, int stroffset, int *lenp) · line 71

Quoted 5 lines
libfdt/fdt_ro.c +1 -1
@@ -68,9 +68,9 @@ const char *fdt_get_string(const void *fdt, int stroffset, int *lenp)
 		}
 	} else if (fdt_magic(fdt) == FDT_SW_MAGIC) {
 		if ((stroffset >= 0)
-		    || (stroffset < -fdt_size_dt_strings(fdt)))
+		    || (stroffset < -(int)fdt_size_dt_strings(fdt)))

So, I'm pretty sure at this point we've tested that fdt_size_dt_strings < INT_MAX. But to make it more obvious that it's safe, I think it would be preferable to case (-stroffset) to unsigned rather than casting size_dt_strings to signed.

Quoted 6 lines
libfdt/fdt_ro.c +1 -1
 			goto fail;
-		if ((-stroffset) < len)
+		if ((unsigned)(-stroffset) < len)
 			len = -stroffset;
 	} else {
 		err = -FDT_ERR_INTERNAL;
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

A message that is not in the archive (sent off-list or not yet received): [email protected]

#47 lore raw

Re: [PATCH 12/14] libfdt: fdt_get_string(): Fix sequential write comparison warnings

Quoted 15 lines · David Gibson
Quoted 14 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in comparisons in fdt_get_string().

The two occassions dealing with the sequential write case are a bit tricky, since we deliberately abuse the negative values here.

As we have just established that stroffset is negative, we can use casts to appease the compiler.

---
 libfdt/fdt_ro.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
Quoted 15 lines

libfdt/fdt_ro.c const char *fdt_get_string(const void *fdt, int stroffset, int *lenp) · line 71

Quoted 10 lines
libfdt/fdt_ro.c +1 -1
index ddb5a2d..059d302 100644
@@ -68,9 +68,9 @@ const char *fdt_get_string(const void *fdt, int stroffset, int *lenp)
 		}
 	} else if (fdt_magic(fdt) == FDT_SW_MAGIC) {
 		if ((stroffset >= 0)
-		    || (stroffset < -fdt_size_dt_strings(fdt)))
+		    || (stroffset < -(int)fdt_size_dt_strings(fdt)))

So, I'm pretty sure at this point we've tested that fdt_size_dt_strings < INT_MAX. But to make it more obvious that it's safe, I think it would be preferable to case (-stroffset) to unsigned rather than casting size_dt_strings to signed.

Ah, indeed, and this makes the whole thing suddenly much more readable. I took the freedom of adding: "unsigned int sw_stroffset = -stroffset;", to avoid those ugly casts altogether.

Cheers,
Andre

Quoted 8 lines
Quoted 7 lines
libfdt/fdt_ro.c +1 -1
 			goto fail;
-		if ((-stroffset) < len)
+		if ((unsigned)(-stroffset) < len)
 			len = -stroffset;
 	} else {
 		err = -FDT_ERR_INTERNAL;
#48 lore raw

PATCH 13/14 libfdt: fdt_node_offset_by_phandle(): Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_node_offset_by_phandle().

Uses a better suited bitwise NOT operator to denote the special value of -1, which automatically results in an unsigned type.

---
 libfdt/fdt_ro.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_ro.c +1 -1
index 059d302..44ad7ec 100644
@@ -680,7 +680,7 @@ int fdt_node_offset_by_phandle(const void *fdt, uint32_t phandle)
 {
 	int offset;
 
-	if ((phandle == 0) || (phandle == -1))
+	if ((phandle == 0) || (phandle == ~0U))
 		return -FDT_ERR_BADPHANDLE;
 
 	FDT_RO_PROBE(fdt);
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-14-andre.przywara-5wv7dgnIgG8@public.gmane.org

Applied
#50 lore raw

Re: [PATCH 13/14] libfdt: fdt_node_offset_by_phandle(): Fix comparison warning

Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_node_offset_by_phandle().

Uses a better suited bitwise NOT operator to denote the special value of -1, which automatically results in an unsigned type.

Applied, thanks.

Quoted 17 lines
---
 libfdt/fdt_ro.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_ro.c +1 -1
index 059d302..44ad7ec 100644
@@ -680,7 +680,7 @@ int fdt_node_offset_by_phandle(const void *fdt, uint32_t phandle)
 {
 	int offset;
 
-	if ((phandle == 0) || (phandle == -1))
+	if ((phandle == 0) || (phandle == ~0U))
 		return -FDT_ERR_BADPHANDLE;
 
 	FDT_RO_PROBE(fdt);
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#51 lore raw

PATCH 14/14 libfdt: fdt_strerror(): Fix comparison warning

In reply to a message not in the archive

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_strerror().

At this point we are sure that errval is negative, and since FDT_ERRTABSIZE is surely positive, we can do some casting to make the compiler happy.

---
 libfdt/fdt_strerror.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_strerror.c +1 -1
index 768db66..efce51d 100644
@@ -48,7 +48,7 @@ const char *fdt_strerror(int errval)
 		return "<valid offset/length>";
 	else if (errval == 0)
 		return "<no error>";
-	else if (errval > -FDT_ERRTABSIZE) {
+	else if (errval > -(int)FDT_ERRTABSIZE) {
 		const char *s = fdt_errtable[-errval].str;
 
 		if (s)
Signature · 1 line
2.17.5

A message that is not in the archive (sent off-list or not yet received): 20200921165303.9115-15-andre.przywara-5wv7dgnIgG8@public.gmane.org

#53 lore raw

Re: [PATCH 14/14] libfdt: fdt_strerror(): Fix comparison warning

Quoted 8 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_strerror().

At this point we are sure that errval is negative, and since FDT_ERRTABSIZE is surely positive, we can do some casting to make the compiler happy.

Since error values are typically conveyed as negative numbers (which is kind of the whole reason we have all these signed values in the first place), I think it would be slightly nicer to actually fold the case to signed into the definition of FDT_ERRTABSIZE.

Quoted 17 lines
---
 libfdt/fdt_strerror.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_strerror.c +1 -1
index 768db66..efce51d 100644
@@ -48,7 +48,7 @@ const char *fdt_strerror(int errval)
 		return "<valid offset/length>";
 	else if (errval == 0)
 		return "<no error>";
-	else if (errval > -FDT_ERRTABSIZE) {
+	else if (errval > -(int)FDT_ERRTABSIZE) {
 		const char *s = fdt_errtable[-errval].str;
 
 		if (s)
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

A message that is not in the archive (sent off-list or not yet received): [email protected]

#55 lore raw

Re: [PATCH 14/14] libfdt: fdt_strerror(): Fix comparison warning

Quoted 8 lines · David Gibson
Quoted 7 lines · Andre Przywara

With -Wsign-compare, compilers warn about a mismatching signedness in a comparison in fdt_strerror().

At this point we are sure that errval is negative, and since FDT_ERRTABSIZE is surely positive, we can do some casting to make the compiler happy.

Quoted 6 lines
Quoted 1 line

Since error values are typically conveyed as negative numbers (which is kind of the whole reason we have all these signed values in the first place), I think it would be slightly nicer to actually fold the case to signed into the definition of FDT_ERRTABSIZE.

Right, makes more sense. I changed the comparison to:
if (-errval < FDT_ERRTABSIZE)
which is more in line with what we actually do below.

Cheers,
Andre

Quoted 18 lines
Quoted 17 lines
---
 libfdt/fdt_strerror.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
libfdt/fdt_strerror.c +1 -1
index 768db66..efce51d 100644
@@ -48,7 +48,7 @@ const char *fdt_strerror(int errval)
 		return "<valid offset/length>";
 	else if (errval == 0)
 		return "<no error>";
-	else if (errval > -FDT_ERRTABSIZE) {
+	else if (errval > -(int)FDT_ERRTABSIZE) {
 		const char *s = fdt_errtable[-errval].str;
 
 		if (s)
#56 lore raw
In reply to a message not in the archive
Quoted 33 lines · Andre Przywara

When libfdt is compiled with -Wsign-compare or -Wextra, GCC emits quite
some warnings about the signedness of the operands not matching:
=================
libfdt/fdt.c:140:18: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
if ((absoffset < offset)
.....
=================

This does not occur under normal conditions in the dtc repo, but might show up when libfdt is embedded in another project. There have been reports from U-Boot and Trusted-Firmware-A.

The underlying issue is mostly due to C's promotion behaviour (ANSI C section 6.1.3.8) when dealing with operands of different signedness (but same size): Signed values get implictly casted to unsigned, which is not typically what we want if they could have been negative.

The Internet(TM) suggests that blindly applying casts is probably doing more harm than it helps, so this series tries to fix the underlying issues properly.
In libfdt, some types are somewhat suboptimal ("int bufsize" comes to mind); some signed types are due to them being returned along wih error values in other functions (node offsets).
So these fixes here have been based on the following assumptions:
- We cannot change the prototype of exported functions.
- It's better to change types (for local variables) than to cast.
- If we have established that a signed value is not negative, we can safely cast it to an unsigned type.

I split up the fixes in small chunks, to make them easier to review. The first four patches change types, the next six ones use casts after we made sure the values are not negative.

Quoted 7 lines

This is only covering libfdt for now (which is what those other projects care about). There are more issues with dtc, but they can be addressed later.

Not sure if some of these checks should be gated by can_assume() calls.

Please have a look, happy to discuss the invididual cases.

Thanks so much for these. I've applied about half of them, the rest I have mostly minor comments for.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

A message that is not in the archive (sent off-list or not yet received): [email protected]

#58 lore raw

On 25/09/2020 05:12, David Gibson wrote:

Hi David,

Quoted 38 lines · Andre Przywara
Quoted 37 lines

When libfdt is compiled with -Wsign-compare or -Wextra, GCC emits quite
some warnings about the signedness of the operands not matching:
=================
libfdt/fdt.c:140:18: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
if ((absoffset < offset)
.....
=================

This does not occur under normal conditions in the dtc repo, but might show up when libfdt is embedded in another project. There have been reports from U-Boot and Trusted-Firmware-A.

The underlying issue is mostly due to C's promotion behaviour (ANSI C section 6.1.3.8) when dealing with operands of different signedness (but same size): Signed values get implictly casted to unsigned, which is not typically what we want if they could have been negative.

The Internet(TM) suggests that blindly applying casts is probably doing more harm than it helps, so this series tries to fix the underlying issues properly.
In libfdt, some types are somewhat suboptimal ("int bufsize" comes to mind); some signed types are due to them being returned along wih error values in other functions (node offsets).
So these fixes here have been based on the following assumptions:
- We cannot change the prototype of exported functions.
- It's better to change types (for local variables) than to cast.
- If we have established that a signed value is not negative, we can safely cast it to an unsigned type.

I split up the fixes in small chunks, to make them easier to review. The first four patches change types, the next six ones use casts after we made sure the values are not negative.

This is only covering libfdt for now (which is what those other projects care about). There are more issues with dtc, but they can be addressed later.

Quoted 6 lines
Quoted 3 lines

Not sure if some of these checks should be gated by can_assume() calls.

Please have a look, happy to discuss the invididual cases.

Thanks so much for these. I've applied about half of them, the rest I have mostly minor comments for.

Many thanks for reviewing (and applying)! I will send the remaining patches ASAP, fixed accordingly.

Cheers,
Andre