Add check for buffer size of Profile.init on 32-bit systems, and fix bug (#41906)

* fix bug multiplying by nthreads twice
* limit buffer to 512MB on 32-bit. Add info to error

Author: IanButterworth

stdlib/Profile/src/Profile.jl

@@ -55,17 +55,26 @@ function init(; n::Union{Nothing,Integer} = nothing, delay::Union{Nothing,Real}
     if n === nothing && delay === nothing
         return Int(n_cur), delay_cur
     end
-    nthreads = Sys.iswindows() ? 1 : Threads.nthreads() # windows only profiles the main thread
-    nnew = (n === nothing) ? n_cur : n * nthreads
+    nnew = (n === nothing) ? n_cur : n
     delaynew = (delay === nothing) ? delay_cur : delay
     init(nnew, delaynew)
 end
 
 function init(n::Integer, delay::Real)
     nthreads = Sys.iswindows() ? 1 : Threads.nthreads() # windows only profiles the main thread
-    status = ccall(:jl_profile_init, Cint, (Csize_t, UInt64), n * nthreads, round(UInt64,10^9*delay))
+    sample_size_bytes = sizeof(Ptr) # == Sys.WORD_SIZE / 8
+    buffer_samples = n * nthreads
+    buffer_size_bytes = buffer_samples * sample_size_bytes
+    if buffer_size_bytes > 2^29 && Sys.WORD_SIZE == 32
+        buffer_size_bytes_per_thread = floor(Int, 2^29 / nthreads)
+        buffer_samples_per_thread = floor(Int, buffer_size_bytes_per_thread / sample_size_bytes)
+        buffer_samples = buffer_samples_per_thread * nthreads
+        buffer_size_bytes = buffer_samples * sample_size_bytes
+        @warn "Requested profile buffer limited to 512MB (n = $buffer_samples_per_thread per thread) given that this system is 32-bit"
+    end
+    status = ccall(:jl_profile_init, Cint, (Csize_t, UInt64), buffer_samples, round(UInt64,10^9*delay))
     if status == -1
-        error("could not allocate space for ", n, " instruction pointers per thread being profiled ($nthreads threads)")
+        error("could not allocate space for ", n, " instruction pointers per thread being profiled ($nthreads threads, $(Base.format_bytes(buffer_size_bytes)) total)")
     end
 end