arch/riscv/mm/init.c:1204:48: sparse: sparse: cast truncates bits from constant value (100000000 becomes 0)

Randy Dunlap rdunlap at infradead.org
Sat Jun 17 16:11:40 PDT 2023


+ Nick
+ linux-riscv

On 5/17/23 03:09, kernel test robot wrote:
> Hi Yimin,
> 
> FYI, the error/warning was bisected to this commit, please ignore it if it's irrelevant.
> 
> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> head:   f1fcbaa18b28dec10281551dfe6ed3a3ed80e3d6
> commit: b5e2c507b06c9d36411845149162a804ae7b04a9 riscv: Kconfig: Allow RV32 to build with no MMU
> date:   7 weeks ago
> config: riscv-randconfig-s041-20230517
> compiler: riscv32-linux-gcc (GCC) 12.1.0
> reproduce:
>         wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
>         chmod +x ~/bin/make.cross
>         # apt-get install sparse
>         # sparse version: v0.6.4-39-gce1a6720-dirty
>         # https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=b5e2c507b06c9d36411845149162a804ae7b04a9
>         git remote add linus https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
>         git fetch --no-tags linus master
>         git checkout b5e2c507b06c9d36411845149162a804ae7b04a9
>         # save the config file
>         mkdir build_dir && cp config build_dir/.config
>         COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=riscv olddefconfig
>         COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=riscv SHELL=/bin/bash arch/riscv/mm/ fs/ext4/
> 
> If you fix the issue, kindly add following tag where applicable
> | Reported-by: kernel test robot <lkp at intel.com>
> | Closes: https://lore.kernel.org/oe-kbuild-all/202305171743.3LP5XWzV-lkp@intel.com/
> 
> sparse warnings: (new ones prefixed by >>)
>    WARNING: invalid argument to '-march': '_zihintpause'
>>> arch/riscv/mm/init.c:1204:48: sparse: sparse: cast truncates bits from constant value (100000000 becomes 0)

^^^ please see below.

> --
>    WARNING: invalid argument to '-march': '_zihintpause'
>    fs/ext4/mballoc.c: note: in included file (through include/linux/timex.h, include/linux/time32.h, include/linux/time.h, ...):
>>> arch/riscv/include/asm/timex.h:25:16: sparse: sparse: cast removes address space '__iomem' of expression
>>> arch/riscv/include/asm/timex.h:25:16: sparse: sparse: incorrect type in argument 1 (different address spaces) @@     expected void const volatile [noderef] __iomem *addr @@     got unsigned int [usertype] * @@
>    arch/riscv/include/asm/timex.h:25:16: sparse:     expected void const volatile [noderef] __iomem *addr
>    arch/riscv/include/asm/timex.h:25:16: sparse:     got unsigned int [usertype] *
>>> arch/riscv/include/asm/timex.h:25:16: sparse: sparse: incorrect type in argument 1 (different address spaces) @@     expected void const volatile [noderef] __iomem *addr @@     got unsigned int [usertype] * @@
>    arch/riscv/include/asm/timex.h:25:16: sparse:     expected void const volatile [noderef] __iomem *addr
>    arch/riscv/include/asm/timex.h:25:16: sparse:     got unsigned int [usertype] *
>    fs/ext4/mballoc.c:2942:13: sparse: sparse: context imbalance in 'ext4_mb_seq_structs_summary_start' - wrong count at exit
> 
> vim +1204 arch/riscv/mm/init.c
> 
> 671f9a3e2e24cd Anup Patel      2019-06-28  1154  
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1155  /*
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1156   * reserve_crashkernel() - reserves memory for crash kernel
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1157   *
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1158   * This function reserves memory area given in "crashkernel=" kernel command
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1159   * line parameter. The memory reserved is used by dump capture kernel when
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1160   * primary kernel is crashing.
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1161   */
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1162  static void __init reserve_crashkernel(void)
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1163  {
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1164  	unsigned long long crash_base = 0;
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1165  	unsigned long long crash_size = 0;
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1166  	unsigned long search_start = memblock_start_of_DRAM();
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1167  	unsigned long search_end = memblock_end_of_DRAM();
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1168  
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1169  	int ret = 0;
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1170  
> d414cb379ac35e Jisheng Zhang   2022-03-23  1171  	if (!IS_ENABLED(CONFIG_KEXEC_CORE))
> d414cb379ac35e Jisheng Zhang   2022-03-23  1172  		return;
> 5640975003d023 Nick Kossifidis 2021-04-19  1173  	/*
> 5640975003d023 Nick Kossifidis 2021-04-19  1174  	 * Don't reserve a region for a crash kernel on a crash kernel
> 5640975003d023 Nick Kossifidis 2021-04-19  1175  	 * since it doesn't make much sense and we have limited memory
> 5640975003d023 Nick Kossifidis 2021-04-19  1176  	 * resources.
> 5640975003d023 Nick Kossifidis 2021-04-19  1177  	 */
> 5640975003d023 Nick Kossifidis 2021-04-19  1178  	if (is_kdump_kernel()) {
> 5640975003d023 Nick Kossifidis 2021-04-19  1179  		pr_info("crashkernel: ignoring reservation request\n");
> 5640975003d023 Nick Kossifidis 2021-04-19  1180  		return;
> 5640975003d023 Nick Kossifidis 2021-04-19  1181  	}
> 5640975003d023 Nick Kossifidis 2021-04-19  1182  
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1183  	ret = parse_crashkernel(boot_command_line, memblock_phys_mem_size(),
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1184  				&crash_size, &crash_base);
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1185  	if (ret || !crash_size)
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1186  		return;
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1187  
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1188  	crash_size = PAGE_ALIGN(crash_size);
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1189  
> a7259df7670240 Mike Rapoport   2021-09-02  1190  	if (crash_base) {
> a7259df7670240 Mike Rapoport   2021-09-02  1191  		search_start = crash_base;
> a7259df7670240 Mike Rapoport   2021-09-02  1192  		search_end = crash_base + crash_size;
> a7259df7670240 Mike Rapoport   2021-09-02  1193  	}
> a7259df7670240 Mike Rapoport   2021-09-02  1194  
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1195  	/*
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1196  	 * Current riscv boot protocol requires 2MB alignment for
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1197  	 * RV64 and 4MB alignment for RV32 (hugepage size)
> decf89f86ecd3c Nick Kossifidis 2021-11-26  1198  	 *
> decf89f86ecd3c Nick Kossifidis 2021-11-26  1199  	 * Try to alloc from 32bit addressible physical memory so that
> decf89f86ecd3c Nick Kossifidis 2021-11-26  1200  	 * swiotlb can work on the crash kernel.
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1201  	 */
> decf89f86ecd3c Nick Kossifidis 2021-11-26  1202  	crash_base = memblock_phys_alloc_range(crash_size, PMD_SIZE,
> decf89f86ecd3c Nick Kossifidis 2021-11-26  1203  					       search_start,
> decf89f86ecd3c Nick Kossifidis 2021-11-26 @1204  					       min(search_end, (unsigned long) SZ_4G));

SZ_4G = 0x100000000, truncated to unsigned long (32 bits), becomes 0, so min() is 0.
Or am I not following this correctly?

Should the second param to min() be changed to SZ_4G - 1
or is this not an issue?

> decf89f86ecd3c Nick Kossifidis 2021-11-26  1205  	if (crash_base == 0) {
> decf89f86ecd3c Nick Kossifidis 2021-11-26  1206  		/* Try again without restricting region to 32bit addressible memory */
> a7259df7670240 Mike Rapoport   2021-09-02  1207  		crash_base = memblock_phys_alloc_range(crash_size, PMD_SIZE,
> a7259df7670240 Mike Rapoport   2021-09-02  1208  						search_start, search_end);
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1209  		if (crash_base == 0) {
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1210  			pr_warn("crashkernel: couldn't allocate %lldKB\n",
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1211  				crash_size >> 10);
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1212  			return;
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1213  		}
> decf89f86ecd3c Nick Kossifidis 2021-11-26  1214  	}
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1215  
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1216  	pr_info("crashkernel: reserved 0x%016llx - 0x%016llx (%lld MB)\n",
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1217  		crash_base, crash_base + crash_size, crash_size >> 20);
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1218  
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1219  	crashk_res.start = crash_base;
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1220  	crashk_res.end = crash_base + crash_size - 1;
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1221  }
> e53d28180d4d0f Nick Kossifidis 2021-04-19  1222  
> 
> :::::: The code at line 1204 was first introduced by commit
> :::::: decf89f86ecd3c3c3de81c562010d5797bea3de1 riscv: try to allocate crashkern region from 32bit addressible memory
> 
> :::::: TO: Nick Kossifidis <mick at ics.forth.gr>
> :::::: CC: Palmer Dabbelt <palmer at rivosinc.com>
> 

Thanks.
-- 
~Randy



More information about the linux-riscv mailing list