address several violations of MISRA Rule 20.7
cover address several violations of MISRA Rule 20.7
Hi all,
this series addresses several violations of Rule 20.7, as well as a small fix to the ECLAIR integration scripts that do not influence the current behaviour, but were mistakenly part of the upstream configuration.
Note that by applying this series the rule has a few leftover violations. Most of those are in x86 code in xen/arch/x86/include/asm/msi.h .
I did send a patch [1] to deal with those, limited only to addressing the MISRA violations, but in the end it was dropped in favour of a more general cleanup of the file upon agreement, so this is why those changes are not included here.
Nicola Vetrini (6):
automation/eclair: address violations of MISRA C Rule 20.7 xen/self-tests: address violations of MISRA rule 20.7 xen/guest_access: address violations of MISRA rule 20.7
x86emul: address violations of MISRA C Rule 20.7 x86/irq: address violations of MISRA C Rule 20.7 automation/eclair_analysis: clean ECLAIR configuration scripts
automation/eclair_analysis/ECLAIR/analyze.sh | 3 +-- automation/eclair_analysis/ECLAIR/deviations.ecl | 8 ++++++++ xen/arch/x86/x86_emulate/x86_emulate.c | 4 ++-- xen/include/xen/guest_access.h | 4 ++-- xen/include/xen/irq.h | 2 +- xen/include/xen/self-tests.h | 8 ++++---- 6 files changed, 18 insertions(+), 11 deletions(-)
Signature · 1 line
2.34.1
PATCH 1/6 automation/eclair: address violations of MISRA C Rule 20.7
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses".
The helper macro bitmap_switch has parameters that cannot be parenthesized in order to comply with the rule, as that would break its functionality. Moreover, the risk of misuse due developer confusion is deemed not substantial enough to warrant a more involved refactor, thus the macro is deviated for this rule.
No functional change.
--- automation/eclair_analysis/ECLAIR/deviations.ecl | 8 ++++++++ 1 file changed, 8 insertions(+)
automation/eclair_analysis/ECLAIR/deviations.ecl +8 -0
@@ -463,6 +463,14 @@ of this macro do not lead to developer confusion, and can thus be deviated." -config=MC3R1.R20.7,reports+={safe, "any_area(any_loc(any_exp(macro(^count_args_$))))"} -doc_end +-doc_begin="The arguments of macro bitmap_switch macro can't be parenthesized as +the rule would require, without breaking the functionality of the macro. This is +a specialized local helper macro only used within the bitmap.h header, so it is +less likely to lead to developer confusion and it is deemed better to deviate it." +-file_tag+={xen_bitmap_h, "^xen/include/xen/bitmap\\.h$"} +-config=MC3R1.R20.7,reports+={safe, "any_area(any_loc(any_exp(macro(loc(file(xen_bitmap_h))&&^bitmap_switch$))))"} +-doc_end + -doc_begin="Uses of variadic macros that have one of their arguments defined as a macro and used within the body for both ordinary parameter expansion and as an operand to the # or ## operators have a behavior that is well-understood and
Signature · 1 line
2.34.1
PATCH 2/6 xen/self-tests: address violations of MISRA rule 20.7
In reply to Nicola VetriniMISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
No functional change.
---
In this case the use of parentheses can detect misuses of the COMPILE_CHECK macro for the fn argument that happen to pass the compile-time check (see e.g. https://godbolt.org/z/n4zTdz595).
An alternative would be to deviate these macros, but since they are used to check the correctness of other code it seemed the better alternative to futher ensure that all usages of the macros are safe.
--- xen/include/xen/self-tests.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
xen/include/xen/self-tests.h +4 -4
@@ -19,11 +19,11 @@#if !defined(CONFIG_CC_IS_CLANG ) ||CONFIG_CLANG_VERSION >=80000 #define COMPILE_CHECK (fn, val, res) \ do { \ - typeof(fn(val)) real = fn(val); \ + typeof((fn)(val)) real = (fn)(val); \ \ if ( !__builtin_constant_p(real) ) \ asm ( ".error \"'" STR(fn(val)) "' not compile-time constant\"" ); \ - else if ( real != res ) \ + else if ( real != (res) ) \ asm ( ".error \"Compile time check '" STR(fn(val) == res) "' failed\"" ); \ } while ( 0 )#else
@@ -37,9 +37,9 @@*/ #define RUNTIME_CHECK (fn, val, res) \ do { \ - typeof(fn(val)) real = fn(HIDE(val)); \ + typeof((fn)(val)) real = (fn)(HIDE(val)); \ \ - if ( real != res ) \ + if ( real != (res) ) \ panic("%s: %s(%s) expected %u, got %u\n", \ __func__, #fn, #val, real, res); \ } while ( 0 )
Signature · 1 line
2.34.1
Re: [XEN PATCH 2/6] xen/self-tests: address violations of MISRA rule 20.7
Quoted 6 lines · Nicola Vetrini
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
Quoted 7 lines
No functional change.
Signed-off-by Nicola Vetrini <[email protected]>---
In this case the use of parentheses can detect misuses of theCOMPILE_CHECKmacro for the fn argument that happen to pass the compile-time check (see e.g. https://godbolt.org/z/n4zTdz595).
While readability suffers a little, I'm okay with the approach taken:
I'd like to give in particular Andrew some time to possibly object, though. And anyway I don't think we want to rush any more Misra changes into 4.19.
Jan
PATCH 3/6 xen/guest_access: address violations of MISRA rule 20.7
In reply to Nicola VetriniMISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
No functional change.
--- xen/include/xen/guest_access.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
xen/include/xen/guest_access.h +2 -2
@@ -49,9 +49,9 @@ ((XEN_GUEST_HANDLE(type)) { &(hnd).p->fld })#define guest_handle_from_ptr (ptr, type) \ - ((XEN_GUEST_HANDLE_PARAM(type)) { (type *)ptr }) + ((XEN_GUEST_HANDLE_PARAM(type)) { (type *)(ptr) })#define const_guest_handle_from_ptr (ptr, type) \ - ((XEN_GUEST_HANDLE_PARAM(const_##type)) { (const type *)ptr }) + ((XEN_GUEST_HANDLE_PARAM(const_##type)) { (const type *)(ptr) })/* * Copy an array of objects to guest context via a guest handle,
Signature · 1 line
2.34.1
Re: [XEN PATCH 3/6] xen/guest_access: address violations of MISRA rule 20.7
Quoted 9 lines · Nicola Vetrini
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
No functional change.
Signed-off-by Nicola Vetrini <[email protected]>
PATCH 4/6 x86emul: address violations of MISRA C Rule 20.7
In reply to Nicola VetriniMISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
No functional change.
---
These local helpers could in principle be deviated, but the readability and functionality are essentially unchanged by complying with the rule, so I decided to modify the macro definition as the preferred option.
--- xen/arch/x86/x86_emulate/x86_emulate.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
xen/arch/x86/x86_emulate/x86_emulate.c +2 -2
@@ -2255,7 +2255,7 @@ x86_emulate(switch ( modrm_reg &7 ) {#define GRP2 (name, ext) \ - case ext: \ + case (ext): \ if ( ops->rmw && dst.type == OP_MEM ) \ state->rmw = rmw_##name; \ else \
@@ -8611,7 +8611,7 @@ int x86_emul_rmw(unsigned long dummy;#define XADD (sz, cst, mod) \ - case sz: \ + case (sz): \ asm ( "" \ COND_LOCK(xadd) " %"#mod"[reg], %[mem]; " \ _POST_EFLAGS("[efl]", "[msk]", "[tmp]") \
Signature · 1 line
2.34.1
Re: [XEN PATCH 4/6] x86emul: address violations of MISRA C Rule 20.7
Quoted 6 lines · Nicola Vetrini
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
Quoted 7 lines
No functional change.
Signed-off-by Nicola Vetrini <[email protected]>---
These local helpers could in principle be deviated, but the readability and functionality are essentially unchanged by complying with the rule, so I decided to modify the macro definition as the preferred option.
Well, yes, but ...
Quoted 12 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1@@ -2255,7 +2255,7 @@ x86_emulate(switch ( modrm_reg &7 ) {#define GRP2 (name, ext) \ - case ext: \ + case (ext): \ if ( ops->rmw && dst.type == OP_MEM ) \ state->rmw = rmw_##name; \ else \
xen/arch/x86/x86_emulate/x86_emulate.c int x86_emul_rmw( · line 8617
Quoted 8 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1@@ -8611,7 +8611,7 @@ int x86_emul_rmw(unsigned long dummy;#define XADD (sz, cst, mod) \ - case sz: \ + case (sz): \ asm ( "" \ COND_LOCK(xadd) " %"#mod"[reg], %[mem]; " \ _POST_EFLAGS("[efl]", "[msk]", "[tmp]") \
... this is really nitpicky of the rule / tool. What halfway realistic ways do you see to actually misuse these macros? What follows the "case" keyword is just an expression; no precedence related issues are possible.
Jan
Re: [XEN PATCH 4/6] x86emul: address violations of MISRA C Rule 20.7
Quoted 31 lines · Jan Beulich
Quoted 6 lines · Nicola Vetrini
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
Quoted 7 lines
No functional change.
Signed-off-by Nicola Vetrini <[email protected]>---
These local helpers could in principle be deviated, but the readability and functionality are essentially unchanged by complying with the rule, so I decided to modify the macro definition as the preferred option.Well, yes, but ...
Quoted 14 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1@@ -2255,7 +2255,7 @@ x86_emulate(switch ( modrm_reg &7 ) {#define GRP2 (name, ext) \ - case ext: \ + case (ext): \ if ( ops->rmw && dst.type == OP_MEM ) \ state->rmw = rmw_##name; \ else \@@ -8611,7 +8611,7 @@ int x86_emul_rmw(unsigned long dummy;
Quoted 11 lines
xen/arch/x86/x86_emulate/x86_emulate.c int x86_emul_rmw( · line 8617
Quoted 6 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1#define XADD (sz, cst, mod) \ - case sz: \ + case (sz): \ asm ( "" \ COND_LOCK(xadd) " %"#mod"[reg], %[mem]; " \ _POST_EFLAGS("[efl]", "[msk]", "[tmp]") \... this is really nitpicky of the rule / tool. What halfway realistic ways do you see to actually misuse these macros? What follows the "case" keyword is just an expression; no precedence related issues are possible.
I do share the view: no real danger is possible in sensible uses. Often MISRA rules are stricter than necessary to have a simple formulation, by avoiding too many special cases.
However, if a deviation is formulated, then it needs to be maintained, for no real readability benefit in this case, in my opinion. I can be convinced otherwise, of course.
Signature · 2 lines
Nicola Vetrini, BSc Software Engineer, BUGSENG srl (https://bugseng.com)
Re: [XEN PATCH 4/6] x86emul: address violations of MISRA C Rule 20.7
Quoted 49 lines · Nicola Vetrini
Quoted 34 lines · Jan Beulich
Quoted 9 lines · Nicola Vetrini
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all
current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.No functional change.
Quoted 7 lines
Signed-off-by Nicola Vetrini <[email protected]>---
These local helpers could in principle be deviated, but the readability
and functionality are essentially unchanged by complying with the rule,
so I decided to modify the macro definition as the preferred option.Well, yes, but ...
Quoted 14 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1@@ -2255,7 +2255,7 @@ x86_emulate(switch ( modrm_reg &7 ) {#define GRP2 (name, ext) \ - case ext: \ + case (ext): \ if ( ops->rmw && dst.type == OP_MEM ) \ state->rmw = rmw_##name; \ else \@@ -8611,7 +8611,7 @@ int x86_emul_rmw(unsigned long dummy;Quoted 13 lines
xen/arch/x86/x86_emulate/x86_emulate.c int x86_emul_rmw( · line 8617
Quoted 6 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1#define XADD (sz, cst, mod) \ - case sz: \ + case (sz): \ asm ( "" \ COND_LOCK(xadd) " %"#mod"[reg], %[mem]; " \ _POST_EFLAGS("[efl]", "[msk]", "[tmp]") \... this is really nitpicky of the rule / tool. What halfway realistic ways do you see to actually misuse these macros? What follows the "case"
keyword is just an expression; no precedence related issues are possible.
Quoted 7 lines
I do share the view: no real danger is possible in sensible uses. Often MISRA rules are stricter than necessary to have a simple formulation, by avoiding too many special cases.
However, if a deviation is formulated, then it needs to be maintained, for no real readability benefit in this case, in my opinion. I can be convinced otherwise, of course.
Well, aiui you're thinking of a per-macro deviation here. Whereas I'd be thinking of deviating the pattern.
Jan
Re: [XEN PATCH 4/6] x86emul: address violations of MISRA C Rule 20.7
Quoted 54 lines · Jan Beulich
Quoted 53 lines · Nicola Vetrini
Quoted 34 lines · Jan Beulich
Quoted 9 lines · Nicola Vetrini
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all
current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.No functional change.
Quoted 7 lines
Signed-off-by Nicola Vetrini <[email protected]>---
These local helpers could in principle be deviated, but the readability
and functionality are essentially unchanged by complying with the rule,
so I decided to modify the macro definition as the preferred option.Well, yes, but ...
Quoted 14 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1@@ -2255,7 +2255,7 @@ x86_emulate(switch ( modrm_reg &7 ) {#define GRP2 (name, ext) \ - case ext: \ + case (ext): \ if ( ops->rmw && dst.type == OP_MEM ) \ state->rmw = rmw_##name; \ else \@@ -8611,7 +8611,7 @@ int x86_emul_rmw(unsigned long dummy;Quoted 13 lines
xen/arch/x86/x86_emulate/x86_emulate.c int x86_emul_rmw( · line 8617
Quoted 6 lines
xen/arch/x86/x86_emulate/x86_emulate.c +1
-1#define XADD (sz, cst, mod) \ - case sz: \ + case (sz): \ asm ( "" \ COND_LOCK(xadd) " %"#mod"[reg], %[mem]; " \ _POST_EFLAGS("[efl]", "[msk]", "[tmp]") \... this is really nitpicky of the rule / tool. What halfway realistic ways do you see to actually misuse these macros? What follows the "case"
keyword is just an expression; no precedence related issues are possible.I do share the view: no real danger is possible in sensible uses. Often MISRA rules are stricter than necessary to have a simple formulation, by avoiding too many special cases.
Quoted 7 lines
Quoted 3 lines
However, if a deviation is formulated, then it needs to be maintained, for no real readability benefit in this case, in my opinion. I can be convinced otherwise, of course.
Well, aiui you're thinking of a per-macro deviation here. Whereas I'd be thinking of deviating the pattern.
Might be reasonable. I'll think about it for v2
Signature · 2 lines
Nicola Vetrini, BSc Software Engineer, BUGSENG srl (https://bugseng.com)
PATCH 5/6 x86/irq: address violations of MISRA C Rule 20.7
In reply to Nicola VetriniMISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
No functional change.
---
Note that the rule does not apply to f because that parameter is not used as an expression in the macro, but rather as an identifier.
--- xen/include/xen/irq.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
xen/include/xen/irq.h +1 -1
@@ -178,7 +178,7 @@ extern struct pirq *pirq_get_info(struct domain *d, int pirq);#define pirq_field (d, p, f, def) ({ \ const struct pirq *__pi = pirq_info(d, p); \ - __pi ? __pi->f : def; \ + __pi ? __pi->f : (def); \ })#define pirq_to_evtchn (d, pirq) pirq_field(d, pirq, evtchn, 0)#define pirq_masked (d, pirq) pirq_field(d, pirq, masked, 0)
Signature · 1 line
2.34.1
Re: [XEN PATCH 5/6] x86/irq: address violations of MISRA C Rule 20.7
Quoted 9 lines · Nicola Vetrini
MISRA C Rule 20.7 states: "Expressions resulting from the expansion of macro parameters shall be enclosed in parentheses". Therefore, some macro definitions should gain additional parentheses to ensure that all current and future users will be safe with respect to expansions that can possibly alter the semantics of the passed-in macro parameter.
No functional change.
Signed-off-by Nicola Vetrini <[email protected]>
PATCH 6/6 automation/eclair_analysis: clean ECLAIR configuration scripts
In reply to Nicola VetriniRemove from the ECLAIR integration scripts an unused option, which was already ignored, and make the help texts consistent with the rest of the scripts.
No functional change.
--- automation/eclair_analysis/ECLAIR/analyze.sh | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
automation/eclair_analysis/ECLAIR/analyze.sh +1 -2
@@ -11,7 +11,7 @@ fatal() { } usage() { - fatal "Usage: ${script_name} <ARM64|X86_64> <Set0|Set1|Set2|Set3>" + fatal "Usage: ${script_name} <ARM64|X86_64> <accepted|monitored>" } if [[ $# -ne 2 ]]; then
@@ -40,7 +40,6 @@ ECLAIR_REPORT_LOG=${ECLAIR_OUTPUT_DIR}/REPORT.log if [[ "$1" = "X86_64" ]]; then export CROSS_COMPILE= export XEN_TARGET_ARCH=x86_64 - EXTRA_ECLAIR_ENV_OPTIONS=-disable=MC3R1.R20.7 elif [[ "$1" = "ARM64" ]]; then export CROSS_COMPILE=aarch64-linux-gnu- export XEN_TARGET_ARCH=arm64
Signature · 1 line
2.34.1