| Message ID | 20220128045004.4843-1-sunilvl@ventanamicro.com |
|---|---|
| State | Accepted |
| Commit | dcf0c838854c86e1f41fb1934aea906845d69782 |
| Headers | show |
| Series | riscv/efi_stub: Fix get_boot_hartid_from_fdt() return value | expand |
Hi Ard, Could you please take this patch? Heinrich and Atish have added RB tag. Let me know if I need to do anything. Thanks Sunil On Fri, Jan 28, 2022 at 10:20:04AM +0530, Sunil V L wrote: > The get_boot_hartid_from_fdt() function currently returns U32_MAX > for failure case which is not correct because U32_MAX is a valid > hartid value. This patch fixes the issue by returning error code. > > Fixes: d7071743db31 ("RISC-V: Add EFI stub support.") > Cc: stable@vger.kernel.org > > Signed-off-by: Sunil V L <sunilvl@ventanamicro.com> > --- > drivers/firmware/efi/libstub/riscv-stub.c | 17 ++++++++++------- > 1 file changed, 10 insertions(+), 7 deletions(-) > > diff --git a/drivers/firmware/efi/libstub/riscv-stub.c b/drivers/firmware/efi/libstub/riscv-stub.c > index 380e4e251399..9c460843442f 100644 > --- a/drivers/firmware/efi/libstub/riscv-stub.c > +++ b/drivers/firmware/efi/libstub/riscv-stub.c > @@ -25,7 +25,7 @@ typedef void __noreturn (*jump_kernel_func)(unsigned int, unsigned long); > > static u32 hartid; > > -static u32 get_boot_hartid_from_fdt(void) > +static int get_boot_hartid_from_fdt(void) > { > const void *fdt; > int chosen_node, len; > @@ -33,23 +33,26 @@ static u32 get_boot_hartid_from_fdt(void) > > fdt = get_efi_config_table(DEVICE_TREE_GUID); > if (!fdt) > - return U32_MAX; > + return -EINVAL; > > chosen_node = fdt_path_offset(fdt, "/chosen"); > if (chosen_node < 0) > - return U32_MAX; > + return -EINVAL; > > prop = fdt_getprop((void *)fdt, chosen_node, "boot-hartid", &len); > if (!prop || len != sizeof(u32)) > - return U32_MAX; > + return -EINVAL; > > - return fdt32_to_cpu(*prop); > + hartid = fdt32_to_cpu(*prop); > + return 0; > } > > efi_status_t check_platform_features(void) > { > - hartid = get_boot_hartid_from_fdt(); > - if (hartid == U32_MAX) { > + int ret; > + > + ret = get_boot_hartid_from_fdt(); > + if (ret) { > efi_err("/chosen/boot-hartid missing or invalid!\n"); > return EFI_UNSUPPORTED; > } > -- > 2.25.1 >
On Jan 28 2022, Sunil V L wrote: > diff --git a/drivers/firmware/efi/libstub/riscv-stub.c b/drivers/firmware/efi/libstub/riscv-stub.c > index 380e4e251399..9c460843442f 100644 > --- a/drivers/firmware/efi/libstub/riscv-stub.c > +++ b/drivers/firmware/efi/libstub/riscv-stub.c > @@ -25,7 +25,7 @@ typedef void __noreturn (*jump_kernel_func)(unsigned int, unsigned long); > > static u32 hartid; > > -static u32 get_boot_hartid_from_fdt(void) > +static int get_boot_hartid_from_fdt(void) I think the function should be renamed, now that it no longer returns the hart ID, but initializes a static variable as a side effect. Thus it no longer "gets", but "sets".
On 2/14/22 10:12, Andreas Schwab wrote: > On Jan 28 2022, Sunil V L wrote: > >> diff --git a/drivers/firmware/efi/libstub/riscv-stub.c b/drivers/firmware/efi/libstub/riscv-stub.c >> index 380e4e251399..9c460843442f 100644 >> --- a/drivers/firmware/efi/libstub/riscv-stub.c >> +++ b/drivers/firmware/efi/libstub/riscv-stub.c >> @@ -25,7 +25,7 @@ typedef void __noreturn (*jump_kernel_func)(unsigned int, unsigned long); >> >> static u32 hartid; >> >> -static u32 get_boot_hartid_from_fdt(void) >> +static int get_boot_hartid_from_fdt(void) > > I think the function should be renamed, now that it no longer returns > the hart ID, but initializes a static variable as a side effect. Thus > it no longer "gets", but "sets". > set_boot_hartid() implies that the caller can change the boot hart ID. As this is not a case this name obviously would be a misnomer. Best regards Heinrich
On Feb 14 2022, Heinrich Schuchardt wrote: > On 2/14/22 10:12, Andreas Schwab wrote: >> On Jan 28 2022, Sunil V L wrote: >> >>> diff --git a/drivers/firmware/efi/libstub/riscv-stub.c b/drivers/firmware/efi/libstub/riscv-stub.c >>> index 380e4e251399..9c460843442f 100644 >>> --- a/drivers/firmware/efi/libstub/riscv-stub.c >>> +++ b/drivers/firmware/efi/libstub/riscv-stub.c >>> @@ -25,7 +25,7 @@ typedef void __noreturn (*jump_kernel_func)(unsigned int, unsigned long); >>> >>> static u32 hartid; >>> >>> -static u32 get_boot_hartid_from_fdt(void) >>> +static int get_boot_hartid_from_fdt(void) >> >> I think the function should be renamed, now that it no longer returns >> the hart ID, but initializes a static variable as a side effect. Thus >> it no longer "gets", but "sets". >> > > set_boot_hartid() implies that the caller can change the boot hart ID. > As this is not a case this name obviously would be a misnomer. Then I guess a different, more fitting name needs to be found.
On Feb 14 2022, Heinrich Schuchardt wrote: > set_boot_hartid() implies that the caller can change the boot hart ID. > As this is not a case this name obviously would be a misnomer. initialize_boot_hartid would fit better.
On 2/14/22 11:15, Andreas Schwab wrote: > On Feb 14 2022, Heinrich Schuchardt wrote: > >> set_boot_hartid() implies that the caller can change the boot hart ID. >> As this is not a case this name obviously would be a misnomer. > > initialize_boot_hartid would fit better. > Another misnomer.
On Feb 14 2022, Heinrich Schuchardt wrote: > On 2/14/22 11:15, Andreas Schwab wrote: >> On Feb 14 2022, Heinrich Schuchardt wrote: >> >>> set_boot_hartid() implies that the caller can change the boot hart ID. >>> As this is not a case this name obviously would be a misnomer. >> >> initialize_boot_hartid would fit better. >> > > Another misnomer. But the best fit so far.
On Mon, Feb 14, 2022 at 12:09:05PM +0100, Andreas Schwab wrote: > On Feb 14 2022, Heinrich Schuchardt wrote: > > > On 2/14/22 11:15, Andreas Schwab wrote: > >> On Feb 14 2022, Heinrich Schuchardt wrote: > >> > >>> set_boot_hartid() implies that the caller can change the boot hart ID. > >>> As this is not a case this name obviously would be a misnomer. > >> > >> initialize_boot_hartid would fit better. > >> > > > > Another misnomer. > > But the best fit so far. Can we use the name init_boot_hartid_from_fdt()? While I understand Heinrich's point, I think since we have "_from_fdt", this may be fine. I didn't rename the function since it was not recommended to do multiple things in a "Fix" patch. If we can consider this as not very serious issue which needs a "Fix" patch, then I can combine this patch with the RISCV_EFI_BOOT_PROTOCOL patch series. Hi Ard, let me know your suggestion on how to proceed with this. Thanks Sunil > > -- > Andreas Schwab, schwab@linux-m68k.org > GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 > "And now for something completely different."
On Thu, Feb 17, 2022 at 2:55 AM Sunil V L <sunilvl@ventanamicro.com> wrote: > > On Mon, Feb 14, 2022 at 12:09:05PM +0100, Andreas Schwab wrote: > > On Feb 14 2022, Heinrich Schuchardt wrote: > > > > > On 2/14/22 11:15, Andreas Schwab wrote: > > >> On Feb 14 2022, Heinrich Schuchardt wrote: > > >> > > >>> set_boot_hartid() implies that the caller can change the boot hart ID. > > >>> As this is not a case this name obviously would be a misnomer. > > >> > > >> initialize_boot_hartid would fit better. > > >> > > > > > > Another misnomer. > > > > But the best fit so far. > > Can we use the name init_boot_hartid_from_fdt()? While I understand > Heinrich's point, I think since we have "_from_fdt", this may be fine. > init_boot_hartid_from_fdt or parse_boot_hartid_from_fdt are definitely much better than the current one. > I didn't rename the function since it was not recommended to do multiple > things in a "Fix" patch. If we can consider this as not very serious > issue which needs a "Fix" patch, then I can combine this patch with the > RISCV_EFI_BOOT_PROTOCOL patch series. > IMHO, it is okay to include this in the RISCV_EFI_BOOT_PROTOCOL series as we are not going to have hartid U32_MAX in a few months :) > Hi Ard, let me know your suggestion on how to proceed with this. > > Thanks > Sunil > > > > -- > > Andreas Schwab, schwab@linux-m68k.org > > GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 > > "And now for something completely different."
On Thu, 17 Feb 2022 at 20:47, Atish Patra <atishp@atishpatra.org> wrote: > > On Thu, Feb 17, 2022 at 2:55 AM Sunil V L <sunilvl@ventanamicro.com> wrote: > > > > On Mon, Feb 14, 2022 at 12:09:05PM +0100, Andreas Schwab wrote: > > > On Feb 14 2022, Heinrich Schuchardt wrote: > > > > > > > On 2/14/22 11:15, Andreas Schwab wrote: > > > >> On Feb 14 2022, Heinrich Schuchardt wrote: > > > >> > > > >>> set_boot_hartid() implies that the caller can change the boot hart ID. > > > >>> As this is not a case this name obviously would be a misnomer. > > > >> > > > >> initialize_boot_hartid would fit better. > > > >> > > > > > > > > Another misnomer. > > > > > > But the best fit so far. > > > > Can we use the name init_boot_hartid_from_fdt()? While I understand > > Heinrich's point, I think since we have "_from_fdt", this may be fine. > > > > init_boot_hartid_from_fdt or parse_boot_hartid_from_fdt > > are definitely much better than the current one. > > > I didn't rename the function since it was not recommended to do multiple > > things in a "Fix" patch. If we can consider this as not very serious > > issue which needs a "Fix" patch, then I can combine this patch with the > > RISCV_EFI_BOOT_PROTOCOL patch series. > > > > IMHO, it is okay to include this in the RISCV_EFI_BOOT_PROTOCOL series > as we are not going to have hartid U32_MAX in a few months :) > > > > Hi Ard, let me know your suggestion on how to proceed with this. > > The patch is fine as it is. I agree that naming is important, but for a helper function that is only used a single time right in the same source file, it doesn't matter that much. I have queued this up now. Thanks, Ard.
diff --git a/drivers/firmware/efi/libstub/riscv-stub.c b/drivers/firmware/efi/libstub/riscv-stub.c index 380e4e251399..9c460843442f 100644 --- a/drivers/firmware/efi/libstub/riscv-stub.c +++ b/drivers/firmware/efi/libstub/riscv-stub.c @@ -25,7 +25,7 @@ typedef void __noreturn (*jump_kernel_func)(unsigned int, unsigned long); static u32 hartid; -static u32 get_boot_hartid_from_fdt(void) +static int get_boot_hartid_from_fdt(void) { const void *fdt; int chosen_node, len; @@ -33,23 +33,26 @@ static u32 get_boot_hartid_from_fdt(void) fdt = get_efi_config_table(DEVICE_TREE_GUID); if (!fdt) - return U32_MAX; + return -EINVAL; chosen_node = fdt_path_offset(fdt, "/chosen"); if (chosen_node < 0) - return U32_MAX; + return -EINVAL; prop = fdt_getprop((void *)fdt, chosen_node, "boot-hartid", &len); if (!prop || len != sizeof(u32)) - return U32_MAX; + return -EINVAL; - return fdt32_to_cpu(*prop); + hartid = fdt32_to_cpu(*prop); + return 0; } efi_status_t check_platform_features(void) { - hartid = get_boot_hartid_from_fdt(); - if (hartid == U32_MAX) { + int ret; + + ret = get_boot_hartid_from_fdt(); + if (ret) { efi_err("/chosen/boot-hartid missing or invalid!\n"); return EFI_UNSUPPORTED; }
The get_boot_hartid_from_fdt() function currently returns U32_MAX for failure case which is not correct because U32_MAX is a valid hartid value. This patch fixes the issue by returning error code. Fixes: d7071743db31 ("RISC-V: Add EFI stub support.") Cc: stable@vger.kernel.org Signed-off-by: Sunil V L <sunilvl@ventanamicro.com> --- drivers/firmware/efi/libstub/riscv-stub.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-)