On Wed, Mar 25, 2026 at 11:29:50PM +0100, Marc Buerg wrote: > proc_do_large_bitmap() does not initialize variable c, which is expected > to be set to a trailing character by proc_get_long(). > > However, proc_get_long() only sets c when the input buffer contains a > trailing character after the parsed value. > > If c is not initialized it may happen to contain a '-'. If this is the > case proc_do_large_bitmap() expects to be able to parse a second part of > the input buffer. If there is no second part an unjustified -EINVAL will > be returned. > > Initialize c to 0 to prevent returning -EINVAL on valid input. > > Fixes: 9f977fb7ae9d ("sysctl: add proc_do_large_bitmap") > Signed-off-by: Marc Buerg > --- > When writing to /proc/sys/net/ipv4/ip_local_reserved_ports it is > possible to receive an -EINVAL for a valid value. > > This happens due to an uninitialized variable in the > proc_do_large_bitmap() function, namely char c. To trigger this behavior > the variable has to contain the later explicitly checked '-' char by > chance. > > In proc_do_large_bitmap() it is expected that the variable might be > filled by the proc_get_long() function with the trailing character of > the given input. But only if a trailing character exists within the > passed size of the buffer. > > The proc_get_long() function can set c if the length of the parsed long > is smaller than the given size of the buffer containing the user input. > This is not the case if the buffer only contains the port value (e.g. > "123") and sets the size exactly to that (3). Meaning if there is no > trailing character, c will not be set. > > If no trailing character is present we still do a c == '-' check. If the > uninitialized variable contains this char the function continues > parsing. It will now set err to -EINVAL in the next proc_get_long() > call, as there is nothing more to parse. > > Initializing c to 0 will solve the problem. > > The problem will only arise sporadically, as the variable must contain > '-' by chance. On the affected system CONFIG_INIT_STACK_NONE=y was > enabled. Further, when enabling eBPF tracing to dump contents of the > stack the issue disappears, which would fit the current explanation as a > root cause for the observed behavior. > --- > Changes in v4: > - Re-include set c to 0 > - Drop check against left > - Move trailers > - Removed Reviewed-by: Peter Seiderer > - Link to v3: https://lore.kernel.org/r/20260319-fix-uninitialized-variable-in-proc_do_large_bitmap-v3-1-9cfc3ff60c09@googlemail.com > > Changes in v3: > - Add Reviewed-by: Peter Seiderer > - Re-include bug context into cover letter > - Link to v2: https://lore.kernel.org/r/20260317-fix-uninitialized-variable-in-proc_do_large_bitmap-v2-1-6dfb1aefa287@googlemail.com > > Changes in v2: > - Drop initialization of c to 0 > - Include checking that left is non-zero before checking against c > - Link to v1: https://lore.kernel.org/r/20260312-fix-uninitialized-variable-in-proc_do_large_bitmap-v1-1-35ad2dddaf21@googlemail.com > --- > kernel/sysctl.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/kernel/sysctl.c b/kernel/sysctl.c > index 9d3a666ffde1..c9efb17cc255 100644 > --- a/kernel/sysctl.c > +++ b/kernel/sysctl.c > @@ -1118,7 +1118,7 @@ int proc_do_large_bitmap(const struct ctl_table *table, int dir, > unsigned long bitmap_len = table->maxlen; > unsigned long *bitmap = *(unsigned long **) table->data; > unsigned long *tmp_bitmap = NULL; > - char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c; > + char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c = 0; > > if (!bitmap || !bitmap_len || !left || (*ppos && SYSCTL_KERN_TO_USER(dir))) { > *lenp = 0; > > --- > base-commit: 80234b5ab240f52fa45d201e899e207b9265ef91 > change-id: 20260312-fix-uninitialized-variable-in-proc_do_large_bitmap-30c6ef4ac1c5 > > Best regards, > -- > Marc Buerg > Thx for the fix Reviewed-by: Joel Granados -- Joel Granados